drones

Review
A Review on Unmanned Aerial Vehicle Remote Sensing:
Platforms, Sensors, Data Processing Methods, and Applications
Zhengxin Zhang 1 and Lixue Zhu 2,3, *

1 College of Information Science and Technology, Zhongkai University of Agriculture and Engineering,
Guangzhou 510225, China; [email protected]
2 Guangdong Laboratory for Lingnan Modern Agriculture, Guangzhou 510642, China
3 School of Mechanical and Electrical Engineering, Zhongkai University of Agriculture and Engineering,
Guangzhou 510225, China
* Correspondence: [email protected]

Abstract: In recent years, UAV remote sensing has gradually attracted the attention of scientific
researchers and industry, due to its broad application prospects. It has been widely used in agriculture,
forestry, mining, and other industries. UAVs can be flexibly equipped with various sensors, such
as optical, infrared, and LIDAR, and become an essential remote sensing observation platform.
Based on UAV remote sensing, researchers can obtain many high-resolution images, with each pixel
being a centimeter or millimeter. The purpose of this paper is to investigate the current applications
of UAV remote sensing, as well as the aircraft platforms, data types, and elements used in each
application category; the data processing methods, etc.; and to study the advantages of the current
application of UAV remote sensing technology, the limitations, and promising directions that still lack
applications. By reviewing the papers published in this field in recent years, we found that the current
application research of UAV remote sensing research can be classified into four categories according
to the application field: (1) Precision agriculture, including crop disease observation, crop yield
estimation, and crop environmental observation; (2) Forestry remote sensing, including forest disease
identification, forest disaster observation, etc.; (3) Remote sensing of power systems; (4) Artificial
facilities and the natural environment. We found that in the papers published in recent years, image
Citation: Zhang, Z.; Zhu, L. A data (RGB, multi-spectral, hyper-spectral) processing mainly used neural network methods; in crop
Review on Unmanned Aerial Vehicle disease monitoring, multi-spectral data are the most studied type of data; for LIDAR data, current
Remote Sensing: Platforms, Sensors, applications still lack an end-to-end neural network processing method; this review examines UAV
Data Processing Methods, and platforms, sensors, and data processing methods, and according to the development process of certain
Applications. Drones 2023, 7, 398. application fields and current implementation limitations, some predictions are made about possible
https://doi.org/10.3390/ future development directions.
drones7060398

Academic Editor: Pablo Keywords: UAV; remote sensing; land applications; UAV imagery
Rodríguez-Gonzálvez

Received: 29 April 2023

Revised: 9 June 2023
1. Introduction
Accepted: 10 June 2023
Published: 15 June 2023 Since the 1960s, Earth observation satellites have garnered significant attention from
both military [1,2] and civilian [3–5] sectors, due to their unique high-altitude observation
ability, enabling simultaneous monitoring of a wide range of ground targets. Since the 1970s,
several countries have launched numerous Earth observation satellites, such as NASA’s
Copyright: © 2023 by the authors. Landsat [6] series; ESA’s SPOT [7] series; and commercial satellites such as IKONOS [8],
Licensee MDPI, Basel, Switzerland. QuickBird, and the WorldView series, generating an enormous volume of remote sensing
This article is an open access article data. These satellites have facilitated the development of several generations of remote
distributed under the terms and sensing image analysis methods, including remote sensing index methods [9–15], object-
conditions of the Creative Commons
oriented analysis methods (OBIA) [16–22], and deep neural network methods [23–27] in
Attribution (CC BY) license (https://
recent years, all of which rely on the multi-spectral and high-resolution images generated
creativecommons.org/licenses/by/
by these remote sensing satellites.
4.0/).

Drones 2023, 7, 398. https://doi.org/10.3390/drones7060398 https://www.mdpi.com/journal/drones

Drones 2023, 7, 398 2 of 42

From the 1980s onward, remote sensing research had mainly been based on satellite
data. Due to the cost of satellite launches, there were only a few remote sensing satellites
available for a long time, and most satellite images required high costs to obtain limited
data, except for a few satellites such as the Landsat series that were partially free. This also
affected the direction of remote sensing research. During this period, many remote sensing
index methods based on ground target spectral characteristics mainly used free Landsat
satellite data. Other satellite data were less used, due to their high purchase costs.
Beside the high cost and lack of supply, remote sensing satellite data acquisition is also
constrained by several factors that affect the observation ability and direction of research:
1. The observation ability of a remote sensing satellite is determined by its cameras. A
satellite can only carry one or two cameras as sensors, and these cameras cannot be
replaced once the satellite has been launched. Therefore, the observation performance
of a satellite cannot be improved in its lifetime;
2. Remote sensing satellites can only observe targets when flying over the adjacent area
above the target and along the satellite’s orbit, which limits the ability to observe
targets from a specific angle;
3. Optical remote sensing satellites use visible and infrared light reflected by observation
targets as a medium, such as panchromatic, colored, multi-spectral, and hyper-spectral
remote sensing satellites. For these satellites, the target illumination conditions se-
riously affect the observation quality. Effective remote sensing imagery data only
can be obtained when the satellite is flying over the observation target and when the
target has good illumination conditions;
4. For optical remote sensing satellites, meteorological conditions, such as cloud cover,
can also affect the observation result, which limits the selection of remote sensing
images for research;
5. The resolution of remote sensing imagery data is limited by the distance between the
satellite and the target. Since remote sensing satellites are far from ground targets,
their image resolution is relatively low.
These constraints not only limit the scope of remote sensing research but also affect
research directions. For instance, land cover/land use is a important aspect of remote
sensing research. However, the research object of land cover/land use is limited by the
spatial resolution of remote sensing image data. The current panchromatic cameras carried
by remote sensing satellites have a resolution of 31 cm/pixel, which can only identify the
type, location, and outline information of ground targets with a 3 m [28] size or more,
such as buildings, roads, trees, ships, cars, etc. Ground objects with smaller sized aerial
projections, such as people, animals, bicycles, etc., cannot be distinguished from the images,
due to the relatively large pixel size. Similarly, change detection, which compares different
information in images taken of the same target in two or more periods, is another example.
Since the data used in many research articles are images taken by the same remote sensing
satellite at different times along its orbit and at the same spatial location, the observation
angles and spatial resolution of these images are similar, making them suitable for pixel-
by-pixel information comparison methods. Hence, change detection has become a key
direction in remote sensing research since the 1980s.
In the past decade, the emergence of multi-rotor unmanned aerial vehicles (UAV) has
gradually changed the above-mentioned limitations in remote sensing research. This type
of unmanned aircraft is pilotless, consumes no fuel, and does not require maintenance of
turboshaft engines. These multi-copters are equipped with cheap but reliable brushless
motors, which only require a small amount of electricity per flight. Users can schedule the
entire flight process of a multi-copter, from takeoff to landing, and edit flight parameters
such as passing points, flight speed, acceleration, and climbing rate. Compared to human-
crewed aircraft such as helicopters and small fixed-wing aircraft, multi-rotor drones are
more stable and reliable, and have several advantages for remote sensing applications.
First, multi-copter drones can carry a variety of sensors flexibly, according to the
requirements of the task. Second, the UAV’s observation angle and target observation time
Drones 2023, 7, 398 3 of 42

are not constrained by specific conditions, as it can be flown by remote control or on a

preset route. Third, even under cloudy, rainy, and night conditions, the UAV can be close to
the target and data can still be obtained. Finally, the spatial resolution of an image obtained
by UAV remote sensing can be up to a millimeter/pixel.
In recent years, there have been several reviews [29–31] on UAV remote sensing.
Some of these reviews [32,33] focused on similar methods developed from satellite remote
sensing in UAV remote sensing data, and some focused on specific application fields, such
as forestry [34,35] and precision agriculture [36–40] remote sensing. In this review, we try to
explore the progress and changes in the application of UAV remote sensing in recent years.
It is worth noting that, besides traditional remote sensing methods such as land cover/land
use and change detection, many recent research papers have employed structure-from-
motion and multi-view stereo (SfM-MVS) methods [41] and LIDAR scanning to obtain
elevation information of ground targets. UAV remote sensing is no longer just a cheap
substitute for satellite remote sensing in industrial and agricultural applications. Instead,
it is now being used to solve problems that were previously difficult to address using
satellite remote sensing, thanks to their flight platform and sensor advantages. As a result,
non-traditional research fields such as forestry, artificial buildings, precision agriculture,
and the natural environment have received increased attention in recent years.
As shown in Figure 1, the structure of this article includes the following sections:
Section 1 is the introduction of the review, which includes the limitations of traditional
satellite remote sensing, the technological background of UAV remote sensing, and the
current application scope. Section 2 introduces the different types of platform and sensor
for drones. Section 3 introduces the processing methods of UAV remote sensing data,
including methods of land cover/land, change detection, and digital elevation models.
Section 4 presents typical application scenarios reflected in recent journal articles on UAV
remote sensing, including forest remote sensing, precision agriculture, power line remote
sensing, artificial targeting, and natural environment remote sensing. Section 5 provides a
discussion, and Section 6 presents the conclusions.

Figure 1. Article organization and content diagram.

2. UAV Platforms and Sensors

The hardware of a UAV remote sensing platform consists of two parts: the flight plat-
form of the drone, and the sensors they are equipped with. Compared to remote sensing
satellites, one of the most significant advantages of UAV remote sensing is the flexible
replacement of sensors, which allows researchers to use the same drone to study the prop-
erties and characteristics of different objects by using different types of sensors. Figure 2
shows this sections’ structure, including the drone’s flight platform and the different types
of sensors carried.
Drones 2023, 7, 398 4 of 42

2.1. UAV Platform

UAVs have been increasingly employed as a remote sensing observation platform
for near-ground applications. Multi-rotor, fixed-wing, hybrid UAVs, and unmanned heli-
copters are the commonly used categories of UAVs. Among these, multi-rotor UAVs have
gained the most popularity, owing to their numerous advantages. These UAVs, which come
in various configurations, such as four-rotor, six-rotor, and eight-rotor, offer high safety
during takeoff and landing and do not require a large airport or runway. They are highly
controllable during flight and can easily adjust their flight altitude and speed. Additionally,
some multi-rotor UAVs are equipped with obstacle detection abilities, allowing them to
stop or bypass obstacles during flight. Figure 3 shows four typical UAV platforms.

Figure 2. UAV platforms and sensors.

Figure 3. UAV platforms: (a) Multi-rotor UAV, (b) Fixed-wing UAV, (c) Unmanned Helicopter,
(d) VTOL UAV.

Multi-rotor UAVs utilize multiple rotating propellers powered by brushless motors to

control lift. This mechanism enables each rotor to independently and frequently adjust its
rotation speed, thereby facilitating quick recovery of flight altitude and attitude in case of
disturbances. However, the power efficiency of multi-rotor UAVs is not prominent, and
their flight duration is relatively short. Common consumer grade drones, after carefully
optimizing their weight and power, have a duration of about 30 min; for example, DJI’s
Mavic Pro has a flight range of 27 min, Mavic 2 has a range of 31 min, and Mavic Air 2 has
a range of 34 min. Despite these limitations, multi-rotor UAVs have been extensively used
as remote sensing data acquisition platforms in the reviewed literature.
Fixed-wing UAVs, which are similar in structure to common aircraft, generate lift
force from the upper and lower air pressure generated by their fixed wings during forward
movement. These UAVs require a runway for takeoff and landing, and their landing
process is more challenging to control than that of multi-rotor UAVs. The stable flight
of fixed-wing UAVs necessitates that the wings provide more lift than the weight of the
aircraft, requiring the UAV to maintain a certain minimum speed throughout its flight.
Consequently, these UAVs cannot hover, and their response to rising or falling airflow is
limited. While the flight speed of fixed-wing UAVs is superior to that of multi-rotor UAVs,
their flight duration is also longer.
Unmanned helicopters, which have a structure similar to helicopters, employ a large
rotor to provide lift and a tail rotor to control direction. These UAVs possess excellent power
efficiency and flight duration, but their mechanical blade structure is complex, leading
to high vibrations and costs. Nonetheless, limited research work on using unmanned
helicopters as a remote sensing platform was reported in the reviewed literature.
Hybrid UAVs, also known as vertical take-off and landing (VTOL), combine the
features of both multi-rotor and fixed-wing UAVs. These UAVs take off and land in multi-
Drones 2023, 7, 398 5 of 42

rotor mode and fly in fixed-wing mode, providing the advantages of easy control during
takeoff and landing and energy-saving during flight.

2.2. Sensors Carried by UAVs

UAVs have been widely utilized as a platform for remote sensing, and the sensors
carried by these aircraft play a critical role in data acquisition. Among the sensors commonly
used by multi-rotor UAVs, there are two main categories: imagery sensors and three-
dimensional information sensors. In addition to the two types of sensor that are commonly
used, other types of sensors carried by drones include gas sensors, air particle sensors,
small radars, etc. Figure 4 shows four typical UAV-carried sensors.

Figure 4. Sensors carried by UAVs :(a) RGB Camera, (b) Multi-spectral Camera, (c) Hyper-spectral
Camera, (d) LIDAR.

Imagery sensors capture images of the observation targets and can be further classified
into several types. RGB cameras capture images in the visible spectrum and are commonly
used for vegetation mapping, land use classification, and environmental monitoring. Multi-
spectral/hyper-spectral cameras capture images in multiple spectral bands, enabling the
identification of specific features such as vegetation species, water quality, and mineral
distribution. Thermal imagers capture infrared radiation emitted by the targets, making it
possible to identify temperature differences and detect heat anomalies. These sensors can
provide high-quality imagery data for various remote sensing applications.
In addition to imagery sensors, multi-rotor UAVs can also carry three-dimensional
information sensors. These sensors are relatively new and have been developed in recent
years with the advancement of simultaneous localization and mapping (SLAM) technology.
LIDAR sensors use laser beams to measure the distance between the UAV and the target,
enabling the creation of high-precision three-dimensional maps. Millimeter wave radar
sensors use electromagnetic waves to measure the distance and velocity of the targets,
making them suitable for applications that require long-range and all-weather sensing.
Multi-camera arrays capture images from different angles, allowing the creation of 3D
models of the observation targets. These sensors can provide rich spatial information,
enabling the analysis of terrain elevation, structure, and volume.

2.2.1. RGB Cameras

RGB cameras are a prevalent remote sensing sensor among UAVs, and two types of
RGB cameras are commonly used on UAV platforms. The first type is the UAV-integrated
camera, which is mounted on the UAV using its gimbal. This camera typically has a
resolution of 20 megapixels or higher, such as the 20-megapixel 4/3-inch image sensor
integrated into the DJI Mavic 3 aircraft and the 20-megapixel 1-inch image sensor integrated
into AUTEL’s EVO II Pro V3 UAV. These cameras can capture high-resolution images at
high frame rates, offering the advantages of being lightweight, compact, and having a long
endurance. However, they cannot replace the original lens with telephoto and wide-angle
lenses, which are required for remote and wide-angle environments.
The second type of camera commonly carried by UAVs is a single lens reflex (SLR)
camera, which enables the replacement of lenses with different focal lengths. UAVs equipped
with SLR cameras offer the advantage of lens flexibility and can be used for remote sensing or
wide-angle observation, making them a valuable tool for such applications. Nonetheless, SLR
Drones 2023, 7, 398 6 of 42

cameras are heavier and require gimbals for installation, necessitating a UAV with sufficient
size and load capacity to accommodate them. For example, Liu et al. [42] utilized the SONY
A7R camera, which provides multiple lens options, including zoom and fixed focus lenses, to
produce a high-precision digital elevation model (DEM) in their research.

2.2.2. Multi-Spectral and Hyper-Spectral Camera

Multi-spectral and hyper-spectral cameras are remote sensing instruments that collect
the spectral radiation intensity of reflected sunlight at specific wavelengths. A multi-
spectral camera is designed to provide data similar to that of multi-spectral remote sensing
satellites, allowing for quantitative observation of the radiation intensity of reflected light
on ground targets in specific sunlight bands. In processing multi-spectral satellite re-
mote sensing image data, the reflected light intensity data of the same ground target in
different spectral bands are used as remote sensing indices, such as the widely used nor-
malized difference vegetation index (NDVI) [9] dimensionless index, which is defined as in
Equation (1):
NIR − Red
NDVI = (1)
NIR + Red
In Equation (1), NIR refers to the measured intensity of reflected light in the near-
infrared spectral range (700∼800 nm), while Red refers to the measured intensity of reflected
light in the red spectral range (600∼700 nm). The NDVI index is used to measure vegetation
density, as living green plants, algae, cyanobacteria, and other photosynthetic autotrophs
absorb red and blue light but reflect near-infrared light. Thus, vegetation-rich areas have
higher NDVI values.
After the launch of the Landsat-1 satellite in 1972, multi-spectral scanner system
(MSS) sensors that can independently observe the ground reflected light according to the
frequency range became a research hot spot data source. When dealing with the problem of
spring vegetation greening and subsequent degradation in the Great Plains of the Central
United States, the studied regional latitude differences are large, so NVDI [9] was proposed
as a spectral index method that is not sensitive to changes of latitude and solar zenith angle.
The NDVI index ranges from 0.3 to 0.8 in densely vegetated areas, and the NDVI value
range is negative for cloud- and snow-covered areas; for a water body, the NDVI value is
close to 0; for bare soil, the NDVI value is a small positive value.
In addition to the vegetation index, other common remote sensing indices include the
normalized difference water index (NDWI) [12], enhanced vegetation index (EVI) [11], leaf
area index (LAI) [43], modified soil adjusted vegetation index (MSAVI) [13], soil adjusted
vegetation index (SAVI) [14], and other remote sensing index methods. These methods
measure the spectral radiation intensity of blue light, green light, red light, red edge,
near-infrared, and other object reflection bands.
Table 1 presents a comparison between the multi-spectral cameras of UAVs and the
multi-spectral sensors of satellites. One notable difference is that a UAV’s multi-spectral
camera has a specific narrow band known as the “red edge” [44], which is not present
in many satellites’ multi-spectral sensors. This band has a wavelength range of 680 nm
to 730 nm, transitioning from the visible light frequencies easily absorbed by plants to
the infrared band largely reflected by plant cells. From a spectral perspective, this band
represents an area where the reflectance of sunlight of plants changes significantly. A few
satellites, such as the European Space Agency(ESA)’s Sentinel-2, have data available in
this band. Research on satellite data has revealed a correlation between leaf area index
(LAI) [43] and this band [45–47]. LAI [43] is a crucial variable in predicting photosynthetic
productivity and evapotranspiration. Another significant difference between UAV multi-
spectral cameras and satellite sensors is the advantage of UAVs’ multi-spectral cameras
in spatial resolution. UAV multi-spectral cameras can reach centimeter/pixel spatial
resolution, which is currently unattainable by satellite sensors. Centimeter-resolution
multi-spectral images have many applications in precision agriculture.
Drones 2023, 7, 398 7 of 42

Table 1. Parameters of UAV multi-spectral cameras and several satellite multi-spectral sensors.

Multi-Spectral Bandwidth and Spatial Resolution

Device Name
Blue Green Red Red Edge Near Infrared I Spatial Resolution
Parrot Sequoia+ None 550 ± 40 nm 1 660 ± 40 nm 735 ± 10 nm 790 ± 40 nm 8 cm/pixel 2
Multi-spectral Rededge-MX 475 ± 32 nm 560 ± 27 nm 668 ± 16 nm 717 ± 12 nm 842 ± 57 nm 8 cm/pixel 2
Camera Altum PT 475 ± 32 nm 560 ± 27 nm 668 ± 16 nm 717 ± 12 nm 842 ± 57 nm 2.5 cm/pixel 2
of UAV Sentera 6X 475 ± 30 nm 550 ± 20 nm 670 ± 30 nm 715 ± 10 nm 840 ± 20 nm 5.2 cm/pixel 2
DJI P4 Multi 3 450 ± 16 nm 560 ± 16 nm 650 ± 16 nm 730 ± 16 nm 840 ± 26 nm
Landsat-5 TM 4 485 ± 35 nm 560 ± 40 nm 660 ± 30 nm None 830 ± 70 nm 30 m/pixel
Landsat-5 MSS 5 None 550 ± 50 nm 650 ± 50 nm None 750 ± 50 nm 60 m/pixel
Multi-spectral Landsat-7 ETM+ 6 485 ± 35 nm 560 ± 40 nm 660 ± 30 nm None 835 ± 65 nm 30 m/pixel
Sensors Landsat-8 OLI 7 480 ± 30 nm 560 ± 30 nm 655 ± 15 nm None 865 ± 15 nm 30 m/pixel
on Satellites IKONOS 8 480 ± 35 nm 550 ± 45 nm 665 ± 33 nm None 805 ± 48 nm 3.28 m/pixel
QuickBird 9 485 ± 35 nm 560 ± 40 nm 660 ± 30 nm None 830 ± 70 nm 2.62 m/pixel
WorldView-4 10 480 ± 30 nm 560 ± 30 nm 673 ± 18 nm None 850 ± 70 nm 1.24 m/pixel
Sentinel-2A 11 492 ± 33 nm 560 ± 18 nm 656 ± 16 nm 745 ± 49 nm 12 833 ± 53 nm 10 m/pixel
1 nm—nanometer. 2 Flight height at 120 m. 3 DJI Phantom 4 Multispectral Camera. 4 Landsat 4–5 Thematic Mapper. 5 Landsat 1–5 Multispectral Scanner. 6 Landsat 7 Enhanced

Thematic Mapper Plus. 7 Landsat 8–9 Operational Land Imager. 8 IKONOS Multispectral Sensor. 9 QuickBird Multispectral Sensor. 10 WorldView-4 Multispectral Sensor. 11 Sentinel-2A
Multispectral Sensor of Sentinel-2. 12 Sentinel-2A have 3 red-edge spectral band, with the spatial resolution 20 m/pixel.
Drones 2023, 7, 398 8 of 42

Hyper-spectral and multi-spectral cameras are both imaging devices that can capture
data across multiple wavelengths of light. However, there are some key differences be-
tween these two types of camera. Multi-spectral cameras typically capture data across a few
discrete wavelength bands, while hyper-spectral cameras capture data across many more
(often hundreds) of narrow and contiguous wavelength bands. Moreover, multi-spectral
cameras generally have a higher spatial resolution than hyper-spectral cameras. Addi-
tionally, hyper-spectral cameras are typically more expensive than multi-spectral cameras.
Table 2 provides a summary of several hyper-spectral cameras and their features and that
were utilized in the papers we reviewed.

Table 2. Parameters of Hyper-spectral Cameras.

Camera Name Spectral Range Spectral Bands Spectral Sampling FWHM 1

Cubert S185 450∼950 nm 125 bands 4 nm 8 nm
Headwall Nano-Hyperspec 400∼1000 nm 301 bands 2 nm 6 nm
RESONON PIKA L 400∼1000 nm 281 bands 2.1 nm 3.3 nm
RESONON PIKA XC2 400∼1000 nm 447 bands 1.3 nm 1.9 nm
HySpex Mjolnir S-620 970∼2500 nm 300 bands 5.1 nm unspecified
1 FWHM–Full Width at Half Maximum of Spectral Resolution.

The data produced by hyper-spectral cameras are not only useful for investigating
the reflected spectral intensity of green plants but also for analyzing the chemical proper-
ties of ground targets. Hyper-spectral data can provide information about the chemical
composition and water content of soil [48], as well as the chemical composition of ground
minerals [49,50]. This is because hyper-spectral cameras can capture data across many
narrow and contiguous wavelength bands, allowing for detailed analysis of the unique
spectral signatures of different materials. The chemical composition and water content of
soil can be determined based on the unique spectral characteristics of certain chemical com-
pounds or water molecules, while the chemical composition of minerals and artifacts can
be identified based on their distinctive spectral features. As such, hyper-spectral cameras
are highly versatile tools that can be utilized for a broad range of applications in various
fields, including agriculture, geology, and archaeology.

2.2.3. LIDAR
LIDAR, an acronym for “laser imaging, detection, and ranging”, is a remote sensing
technology that has become increasingly popular in recent years, due to its ability to
generate precise and highly accurate 3D images of the Earth’s surface. LIDAR systems
mounted on UAVs are capable of collecting data for a wide range of applications, including
surveying [51,52], environmental monitoring [53], and infrastructure inspection [54–56].
One of the key advantages of using LIDAR in UAV remote sensing is its ability to
provide highly accurate and detailed elevation data. By measuring the time it takes for
laser pulses to bounce off the ground and return to the sensor, LIDAR can create a high-
resolution digital elevation model (DEM) of the terrain. This data can be used to create
detailed 3D maps of the landscape, which are useful for a variety of applications, such as
flood modeling, land use planning, and urban design.
Another benefit of using LIDAR in UAV remote sensing is its ability to penetrate
vegetation cover to some extent, allowing for the creation of detailed 3D models of forests
and other vegetation types. Multiple return LIDAR has the ability to measure the return
time of different pulses of reflected light emitted at the same time. By precisely using this
feature, information on the canopy structure in a forest can be obtained by measuring the
different return times. This data can be used for ecosystem monitoring, wildlife habitat
assessment, and other environmental applications.
Drones 2023, 7, 398 9 of 42

In addition to mapping and environmental monitoring, LIDAR-equipped UAVs are

also used for infrastructure inspection and construction environment monitoring. By
collecting high-resolution images of bridges, buildings, and other structures, LIDAR can
help engineers and construction professionals identify potential problems. Figure 5 shows
mechanical scanning and solid-state LIDAR.

Figure 5. LIDAR: (a) Mechanical Scanning LIDAR, (b) Solid-state LIDAR.

LIDAR technology has evolved significantly in recent years with the emergence of
solid-state LIDAR technology, which uses an array of stationary lasers and photodetectors
to scan the target area. Solid-state LIDAR technology offers several advantages over
mechanical scanning LIDAR, which use a rotating mirror or prism to scan a laser beam
across the target area. Solid-state LIDAR is typically more compact and lightweight, making
it well suited for use on UAVs.

3. UAV Remote Sensing Data Processing

UAV remote sensing has several advantages compared with satellite remote sensing:
(1) UAV remote sensing can be equipped with specific sensors for observation, as required.
(2) UAV remote sensing can observe targets at any time period allowed by weather and
environmental conditions. (3) UAV remote sensing can set a repeatable flight route, to
achieve multiple target observations from a set altitude and angle. (4) The image sensor
mounted on the UAV is closer to the target, and the image resolution obtained by observa-
tion is higher. These characteristics have not only allowed the remote sensing community
to produce new techniques in land cover/land use and change detection based on remote
sensing satellite data in the past, but have also contributed to the growth of forest remote
sensing, precision agriculture remote sensing, and other research directions.

3.1. Land Cover/Land Use

Land cover and land use are fundamental topics in satellite remote sensing research.
This field aims to extract information about ground observation targets from low-resolution
image data captured by early remote sensing satellites. NASA’s Landsat series satellite
program is the longest-running Earth resource observation satellite program to date, with
50 years of operation since the launch of Landsat-1 [57] in 1972.
In the early days of remote sensing, land use classification methods focused on identi-
fying and classifying the spectral information of pixels covering the target object, known as
sub-pixel approaches [58]. The concept of these methods is that the spectral characteristics
of a single pixel in a remote sensing image are based on the spatial average of the spectral
signatures reflected from multiple object surfaces within the area covered by that pixel.
However, with the emergence of high-resolution satellites, such as QuickBird and
IKONOS, which can capture images with meter-level or decimeter-level spatial resolution,
the industry has produced a large amount of high-resolution remote sensing data with
Drones 2023, 7, 398 10 of 42

sufficient object textural features. This has led to the development of object-based image
analysis (OBIA) methods for land use/land cover.
OBIA uses a super-pixel segmentation method to segment the image and then applies
a classifier method to classify the spectral features of the segmented blocks and identify
the type of ground targets. In recent years, neural network methods, especially the full
convolution neural network (FCN) [59] method, have become the mainstream methods of
land use and land cover research. Semantic segmentation [23,60,61] and instance segmen-
tation [24,62,63] neural network methods can extract the type, location, and spatial range
information of ground targets end-to-end from remote sensing images.
The emergence of unmanned aerial vehicle (UAV) remote sensing has produced a
new generation of data for land cover/land use research. The image sensors carried by
UAVs can acquire images with decimeter-level, centimeter-level, or even millimeter-level
resolution, allowing the problem of information extraction for small objects on the ground,
which were previously difficult to study, to become a new research interest, such as people
on the street, cars, animals, and plants.
Researchers have proposed various methods to address these challenges. For instance,
PEMCNet [64], an encoder–decoder neural network method proposed by Zhao et al.,
achieved good classification results for LIDAR data taken by UAVs, with a high accuracy
for ground objects such as buildings, shrubs, and trees. Harvey et al. [65] proposed a
terrain matching system based on the Xception [66] network model, which uses a pre-
trained neural network to determine the position of the aircraft without relying on inertial
measurement units (IMUs) and global navigation satellite systems (GNSS). Additionally,
Zhuang et al. [67] proposed a method based on neural networks to match remote sens-
ing images of the same location taken from different perspectives and resolutions, called
multiscale block attention (MSBA). By segmenting and combining the target image and
calculating the loss function separately for the local area of the image, the authors realized
a matching method for complex building targets photographed from different angles.

3.2. Change Detection

Remote sensing satellites can observe the same target area multiple times. Comparing
the images obtained from two observations, we can detect changes in the target area over
time. Change detection using remote sensing satellite data has wide-ranging applications,
such as in urban planning, agricultural surveying, disaster detection and assessment, map
compilation, and more.
UAV remote sensing technology allows for data acquisition from multiple aerial
photos taken at different times along a preset route. Compared to other types of remote
sensing, UAV remote sensing has advantages in spatial resolution and data acquisition for
change detection. Some of the key benefits include: (1) UAV remote sensing operates at
a lower altitude, making it less susceptible to meteorological conditions such as clouds
and rain. (2) The data obtained through UAV remote sensing are generated through
structure-from-motion and multi-view stereo (SfM-MVS) and airborne laser scanning (ALS)
methods, which enable the creation of a DEM for the observed target and adjacent areas,
allowing us to monitor changes in three dimensions over time. (3) UAVs can acquire data
at user-defined time intervals by conducting multiple flights in a short time.
Recent research on change detection based on UAV remote sensing data has focused
on identifying small and micro-targets, such as vehicles, bicycles, motorcycles, and tricycles,
and tracking their movements using UAV aerial images and video data. Another area of
research involves the practical application of UAV remote sensing for detecting changes in
3D models of terrain, landforms, and buildings.
For instance, Chen et al. [68] proposed a method to detect changes in buildings using
RGB images obtained from UAV aerial photography and 3D reconstruction of RGB-D
data. Cook et al. [69] compared the accuracy of 3D models generated using a SfM-MVS
method and LIDAR scanning measurement for reconstructing complex mountainous river
terrain areas, with a root-mean-square error (RMSE) of 30∼40 cm. Mesquita et al. [70]
Drones 2023, 7, 398 11 of 42

developed a change detection method, which was tested on the Oil Pipes Construction
Dataset(OPCD) and successfully detected construction traces from multiple pictures taken
by UAV at different times in the same area and space. Hastaouglu et al. [71] monitored
three-dimensional displacement in a garbage dump using aerial image data and the SfM-
MVS method [41] to generate a three-dimensional model. Lucieer et al. [72] proposed a
method for reconstructing a three-dimensional model of landslides in mountainous areas
based on unmanned aerial vehicle multi-view images using the SfM-MVS method. The
measured horizontal accuracy was 7 cm, and the vertical accuracy was 6 cm. Li et al. [73]
monitored the deformation of the slope section of large water conservancy projects using
UAV aerial photography and achieved a measurement error of less than 3 mm, which was
significantly higher than traditional aerial photography methods. Han et al. [74] proposed
a method of using UAVs to monitor road construction, which was applied to an extended
road construction site and accurately identified changed ground areas with an accuracy
of 84.5∼85%. Huang et al. [75] developed a semantic detection method for changes in
construction sites, based on a 3D point cloud data model generated from images obtained
through UAV aerial photography.

3.3. Digital Elevation Model (DEM) Information

In recent years, the accurate generation of digital elevation models (DEM) has become
increasingly important in remote sensing landform research. DEMs provide crucial infor-
mation about ground elevation, including both digital terrain models (DTM) and digital
surface models (DSM). A DTM represents the natural surface elevation, while a DSM
includes additional features such as vegetation and artificial objects. There are two primary
methods for calculating elevation information: structure-from-motion and multi-view
stereo (SfM-MVS) [41] and airborne laser scanning (ALS).
Among the reviewed articles, the SfM-MVS method gained more attention due to
its simple requirements. Sanz-Ablanedo et al. [76] conducted a comparative experiment
to assess the accuracy of the SfM-MVS method when establishing a DTM model in a
complex mining area covering over 1200 hectares (1.2 × 107 m2 ). The results showed that
when a small number of ground control points (GCPs) were used, the root-mean-square
error (RMSE) of the checkpoint was plus or minus five times the ground sample distance
(GSD), or about 34 cm. In contrast, when more GCPs were used (i.e., more than 2 GCP in
100 images), the RMSE of the checkpoint response converged to twice the GSD, or about
13.5 cm. Increasing the number of GCPs had a significant impact on the accuracy of the
3D-model generated by the SfM-MVS method. It is worth noting that the authors used
a small fixed-wing UAV as their remote sensing platform. Rebelo et al. [77] proposed a
method to generate a DTM by taking RGB images from multi-rotor UAVs. The authors
used an RGB sensor carried by a DJI Phantom 4 UAV to take images within an area of
55 hectares (5.5 × 105 m2 ) and established a 3D point cloud DTM through the SfM-MVS
method. Although the GNSS receiver used was the same model, the horizontal RMSE of the
DTM was 3.1 cm, the vertical RMSE was 8.3 cm, and the comprehensive RMSE was 8.8 cm.
This precision was much better than that of the fixed-wing UAV method of Sanz-Ablanedo
et al. [76]. In another study, Almeida et al. [78] proposed a method for qualitative detection
of single trees in forest land based on UAV remote sensing RGB data. In their experiment,
the authors used a 20-megapixel camera carried by a DJI Phantom 4 PRO to reconstruct
a DTM in the SfM-MVS mode of Agisoft Metashape, over an area of 0.15 hectares. For
the DTM model obtained, the RMSE of GCPs in the horizontal direction was 1.6 cm, and
that in the vertical direction was 3 cm. Hartwig et al. [79] reconstructed different forms of
ravine using SfM-MVS based on multi-view images captured by multiple drones. Through
experiments, the authors verified that, even without using GCP for geographic registration,
SfM-MVS technology alone could achieve a 5% accuracy in the volume measurement
of ravines.
In airborne laser scanning (ALS) methods, Zhang et al. [53] proposed a method to
detect ground height in tropical rainforests based on LIDAR data. This method involved
Drones 2023, 7, 398 12 of 42

scanning a forest area with airborne LIDAR to obtain three-dimensional point cloud data.
Local minima were extracted from the point cloud data as candidate points, with some
of these candidates representing the ground between trees in the forest area. The DTM
generated by the method had high consistency with the ALS-based reference, with a RMSE
of 2.1 m.

4. UAV Remote Sensing Application

In recent years, the utilization of UAV remote sensing technology has gained significant
attention in a variety of fields, such as agriculture [80–83], forestry [84–86], environmental
monitoring [87–89], and infrastructure inspection [55,67,90]. UAVs provide high-resolution
and accurate data that can facilitate informed decision-making and support diverse appli-
cations. With the continuous advancement of UAV technology, we can anticipate further
innovative and impactful applications in the foreseeable future. Figure 6 shows the organi-
zation of the sections on UAV applications.

Figure 6. UAV remote sensing applications.

4.1. UAV Remote Sensing in Forestry

Forestry remote sensing is a relatively small aspect of satellite remote sensing. Multi-
spectral and hyper-spectral data related to tree photosynthesis are among the various
types of remote sensing satellite data. However, the spatial resolution of remote sensing
satellite data is limited and cannot meet the requirements for specific tree types and
disease identification. Additionally, remote sensing satellites cannot provide the high-
accuracy elevation information necessary for forestry structure studies. Figure 7 shows the
organization of the applications in the forestry section.

Figure 7. UAV remote sensing research in forestry.

Drones 2023, 7, 398 13 of 42

UAV-based remote sensing technology has introduced a new dimension to forestry

remote sensing. With the ability to carry high-resolution and multi-spectral cameras, UAV
remote sensing can identify tree types and observe forest diseases. It can also use LIDAR
to measure the canopy information of multi-layered forest canopies. Therefore, in recent
years, UAV remote sensing technology has emerged as a developing research direction
for monitoring forestry ecology. Its primary research areas include (1) Estimation of forest
structure parameters; (2) Classification and identification of forest plants; (3) Monitoring of
forest fires; and (4) Monitoring of forest diseases.

4.1.1. Estimation of Forest Structural Parameters

The estimation of forest structural parameters, including tree height and canopy
parameters, is a crucial research area in forestry. UAV remote sensing technology provides
an efficient and accurate approach for estimating these parameters.
Krause et al. [51] investigated two different methods of measuring tree height in
dense forests: field measurement, and a method using UAV remote sensing RGB image
data. The authors applied UAV remote sensing to measure multiple trees and obtained
a RMSE = 0.479 m (2.78%). Ganz et al. [91] investigated the difference in measured tree
heights between LiDAR data and UAV RGB images. The authors achieved a RMSE = 0.36 m
based on LIDAR data, with a RMSE = 2.89 m on RGB image. Fakhri et al. [92] proposed an
object-based image analysis (OBIA) method to estimate tree height and canopy area. The
authors employed marker-controlled watershed segmentation (MCWS) [93] to segment the
UAV aerial images and classify and post-process the ground target with information from
a digital surface model (DSM) and digital terrain model (DTM).
Pu et al. [94] proposed two methods to measure the canopy closure (CC) parameters of
trees using unmanned aerial LIDAR data, which can replace the time-consuming and labo-
rious hemispheric photography (HP) method. The first method is based on a canopy-height
model (CHM), while the second method is based on synthetic hemispheric photography
(SHP). The CHM-based method exhibited a high accuracy in the range of 45 degrees zenith
angle, but the accuracy decreased rapidly in the range of 60 to 75 degrees zenith angle. In
contrast, the SHP-based method demonstrated a high accuracy in the range of 45, 60, and
75 degrees.
Zhang et al. [53] proposed a method to detect the ground height of tropical rainforest
using LIDAR data. The method has three steps: (1) Selecting points with a local minimum
value from a large number of point cloud data as candidate points; (2) Comparing each
point in the selected local minimum point set with the DTM of their sampling source
location, and the points with errors within 2 m are considered as real ground points, while
the rest are regarded as non-ground points; and (3) Training a random forest classifier
using the local minimum point set of marker type as the training data. This classifier can
be used to distinguish ground points from other regional point cloud data. Finally, the
DTM is interpolated from the ground points obtained by the random forest classifier using
a downsampling method.

4.1.2. Classification and Identification of Forest Plants

Classification and identification of forest plants is an important application of UAV
remote sensing in forestry. Different methods have been proposed to achieve this goal.
Mo et al. [95] proposed a neural network method for litchi tree instance segmentation
using multi-spectral remote sensing image data. Reder et al. [96] proposed a semantic
segmentation neural network method to monitor collapsed trees in a forest after a storm
using UAV data. Yu et al. [52] developed a method to classify forest vertical structure
using spectral index maps generated from multi-spectral data and canopy height maps
generated from LIDAR data. They compared the classification results of three classifiers:
random forest (RF), XGBoost [97], and support vector machine (SVM), and obtained the
best results with XGBoost. Guo et al. [98] proposed a method to extract tree seedlings in a
cost-effective manner in a complex vegetation environment using image data collected by
Drones 2023, 7, 398 14 of 42

UAV. They combined RGB and grey-level co-occurrence matrix (GLCM) features and used
a random forest classifier to identify the crown area in the image. Taylor-Zavala et al. [99]
investigated the correlation between the biochemical characteristics of plant cells and their
spectral characteristics by comparing the multi-spectral data collected by UAV with the
biochemical characteristics of sampled plant leaves.

4.1.3. Monitoring of Forest Fires

Fire is one of the most severe disasters in forest areas, with the potential to spread
rapidly and cause widespread damage. Given the potential threat to the safety of first-
line firefighters, it is important to have an effective and safe method of monitoring and
evaluating forest fires. UAVs are suitable tools for forest fire monitoring, given their ability
to provide remote sensing information of fire in real time.
Li et al. [100] proposed a method for predicting the spread process of forest fires using
a long–short-term memory (LSTM) [101] neural network model called FNU-LSTM. The
authors collected video data of the forest fire spread process using an infrared camera
mounted on a UAV and trained the LSTM network model to predict the forest fire spread
rate. Hu et al. [102] developed a method for monitoring forest fires using a group of UAVs
for remote sensing. Namburu et al. [103] proposed a method to identify forest fires using
UAV remote sensing RGB image data. The authors trained an improved artificial neural
network, x-mobilenet, a varied MobileNet [104] network model to identify forest fires
through the RGB image data of forest fires collected by UAVs and an existing public fire
image database, achieving an accuracy rate of 97.22%.
Beltrán-Marcos et al. [105] investigated the relationship between satellite and UAV
remote sensing multi-spectral image data for measuring fire severity after a forest fire. The
authors found that soil organic carbon (SOC) and soil water content (SMC), which can
be estimated using multi-spectral image data, were highly correlated with the severity of
the fire.

4.1.4. Monitoring of Forest Diseases

Monitoring forest diseases is essential for forest protection. Pine wilt is a widespread
tree disease, with a concentrated research field. The reasons for this include its obvious
external symptoms, which can be observed and detected from RGB, multi-spectral, and
hyper-spectral image data. Pine is a widely distributed tree species in wild forest areas.
Pine wilt disease is widely distributed in forest areas with average temperatures exceed-
ing 20 degrees Celsius [106] and is considered a global threat to forest ecosystems [107].
Table 3 lists part of the reviewed forestry disease articles and compares their data collection
locations, drone platforms, sensors, tree species, disease types, and methods of forestry
disease identification. Figure 8 shows symptoms of pine wilt disease.

Figure 8. Symptoms of pine wilt disease.

Hu et al. [86] proposed a neural network method based on UAV remote sensing
RGB image data to identify diseased pine trees in a forest. The spatial resolution of the
dataset sampled by the authors was 5 cm/pixel, and the UAV’s flight height was 380 m.
The authors tested several classifiers on the sampled data, and the proposed deep neural
network achieved the highest recall rate (91.3%). Wu et al. [108] proposed a method to
identify pine wilt based on UAV remote sensing RGB image data. The author divided the
Drones 2023, 7, 398 15 of 42

disease course of pine wilt into early stage and late stage, plus healthy pine, and divided
this into three categories. In the experiment, the recognition accuracy of the neural network
method for late stage of fusarium wilt (73.9∼77.2%) was much higher than that of the early
stage of fusarium wilt (46.5∼50.8%). Xia et al. [109] studied the shape of pine blight based
on camera RGB data taken by a UAV platform and combined with a ground survey. Their
research showed that from the RGB images taken with the ordinary SLR camera, using
current neural network tools, the accuracy of detecting pine blight could reach 80.6%, while
the recall rate could reach 83.1%. Li et al. [110] proposed a pine wilt detection method that
can be used with edge computing hardware and that can be placed on UAVs. This method
is based on remote sensing RGB data from a UAV and uses the YOLOv4 [111] tiny neural
network model to detect the infection of pine with fusarium wilt. Ren et al. [84] proposed
a neural network method for detecting pine wilt from UAV remote sensing RGB image
data. The spatial resolution of the RGB image sampled by the authors was 5 cm/pixel. The
authors considered diseased pine trees as positive samples and other red ground targets
(such as red cars, red roofs, and red floors) as negative samples. All samples were checked
with boxes. The recall rate of this method was 86.6%, and the accuracy rate was 79.8%.
Sun et al. [112] proposed an object-oriented method for detecting pine wilt using UAV
remote sensing RGB data.
Yu et al. [85] proposed a neural network method for identifying pine wilt based on
UAV remote sensing multi-spectral data. The UAV flew at a height of 100 m, and the spatial
resolution of multi-spectral images was 12 cm/pixel. The authors divided the course of
pine wilt disease into three stages: early stage, middle stage, and late stage. In addition,
healthy pines and other broad-leaved trees were added as two kinds of comparison samples,
which made a total of five types of sample. Based on the classified multi-spectral data, the
authors found that the correct recognition rate was nearly 50% in the early stage of pine
wilt and more than 70% in the middle stage of pine wilt. Yu et al. [113] also proposed a
method to detect early pine wilt disease based on UAV hyper-spectral image data. The
authors ran a UAV at a flight height of 120 m above ground, using a Resonon Pika L hyper-
spectral camera for sampling data. The spatial resolution of the image was 44 cm/pixel,
and the LR1601-IRIS LIDAR system was used to collect LIDAR point cloud data. The
authors classified UAV hyper-spectral forest image data using a 3D convolution neural
network, and the comprehensive accuracy rate of the five types of ground target (pine
wilt at early stage, middle stage and late stage, health pine, and other broad-leaved trees)
reached 88.11%. Yu et al. [114] also proposed a method to identify pine wilt based on
UAV hyper-spectral and LIDAR data. The UAV flew at a height of 70 m, and the spatial
resolution of hyper-spectral images was 25.6 cm/pixel. Using a random forest classifier,
the authors recognized five types ground target (pine wilt at early stage, middle stage and
late stage, health pine, and other broad-leaved trees). Under the condition of using only
hyper-spectral data, the classification accuracy was 66.86%; under the condition of using
only LIDAR data, the accuracy was 45.56%; with the combination of the two data sources,
the accuracy reached 73.96%. Li et al. [115] proposed a method of data recognition based
on UAV remote sensing hyper-spectral images. Using a Headwall Nano-Hyperspec as an
instrument, 8 data blocks were obtained, each a size of 4600 × 700 pixels, and with a spatial
resolution of 11 cm/pixel. When tested on these different data blocks, the accuracy rate of
this method was from 84% to 99.8%, and the recall rate was from 88.3% to 99.9%.
For other types of trees, Dash et al. [116] simulated the changes of leaf reflectance
spectrum characteristics caused by forest disease outbreaks through small-scale experi-
ments. This result also proved the feasibility of monitoring forest diseases through UAV
multi-spectral remote sensing. Sandino et al. [117] proposed a method for detecting fungal
infection of trees in forests based on UAV remote sensing hyper-spectral image data. The
authors took images of paperbark tea trees from a forest environment under the condition
of partial myrtle rust infection using a Headwall Nano-Hyperspec hyper-spectral camera.
The image marked each pixel according to five types: health, infected, background, soil,
and protruding tree stump. By training the XGBoost [97] classifier, the authors obtained
Drones 2023, 7, 398 16 of 42

a comprehensive accuracy of 97.35% on the validation data. Nasi et al. [118] proposed a
method for detecting the damage of spruce from beetle disease in a forest based on the
UAV remote sensing hyper-spectral data. Gobbi et al. [119] experimented with a method of
identifying the degraded state of forests using UAV remote sensing RGB image data. The
proposed method generates a 3D point cloud and canopy height model (CHM) through
the SfM-MVS method based on UAV remote sensing RGB images. The method measures
forest degradation from three dimensions: forest structural integrity, structural complexity,
and forest middle vegetation density. The authors stated that the SfM-MVS method is an
appropriate tool for building and evaluating forest degradation models. Coletta et al. [120]
proposed a method for identifying eucalyptus infection with ceratocystis wilt disease based
on UAV remote sensing RGB image data. Xiao et al. [121] proposed a method for detecting
apple tree fire blight based on UAV remote sensing multi-spectral data.

4.2. Remote Sensing for Precision Agriculture

Precision agriculture [122,123] can be defined as the application of modern informa-
tion technology to process multi-source data for decision-making and operations in crop
production management.
The utilization of UAV remote sensing allows for the acquisition of high-resolution
imagery data that facilitates detailed observations of agricultural disease. This approach
has gained popularity among researchers in recent years, due to its ability to enhance
crop disease observation, water management, weed management, crop monitoring, and
yield estimation. UAV remote sensing has become an increasingly useful tool for pre-
cision agriculture applications. Figure 9 shows the organization of the applications in
precision agriculture.

Figure 9. UAV remote sensing research in precision agriculture.

Drones 2023, 7, 398 17 of 42

Table 3. Papers reviewed in forest diseases.

Year Authors Study Area UAV Type Sensor Species Disease Method Type
2020 Hu et al. [86] Anhui, China Mulit-rotor RGB Pine Pine Wilt Neural Network
2021 Wu et al. [108] Qingkou, Fujian, China Mulit-rotor RGB Pine Pine Wilt Neural Network
2021 Xia et al. [109] Qingdao, Shandong, China Fixed-wing RGB Pine Pine Wilt Neural Network
2021 Li et al. [110] Taián, Shandong, China Mulit-rotor RGB Pine Pine Wilt Neural Network
2022 Ren et al. [84] Yichang, Hubei, China Mulit-rotor RGB Pine Pine Wilt Neural Network
2022 Sun et al. [112] Dayu, Jiangxi, China Mulit-rotor RGB Pine Pine Wilt OBIA
2021 Yu et al. [85] Yiwu, Zhejiang, China Mulit-rotor Multi-spectral Pine Pine Wilt Neural Network
2021 Yu et al. [113] Fushun, Liaoning, China Mulit-rotor Hyper-spectral Pine Pine Wilt Neural Network
2021 Yu et al. [114] Yiwu, Zhejiang, China Mulit-rotor Hyper-spectral & LiDAR Pine Pine Wilt Random Forest
2022 Li et al. [115] Yantai, Shandong, China Mulit-rotor Hyper-spectral Pine Pine Wilt Neural Network
2022 Coletta et al. [120] Fixed-wing RGB Eucalyptus Ceratocystis Wilt Ensemble Method
2022 Xiao et al. [121] Biglerville, Pennsylvania, USA Mulit-rotor Multi-spectral Apple Apple Fire Blight Vegetation Index
Drones 2023, 7, 398 18 of 42

4.2.1. Crop Disease Observation

Crop diseases can be caused by a variety of pathogens, including viruses, fungi,
bacteria, and insects, which can cause phenotypic changes in different plants. As the
diameter of the spots on the leaves of plants in the early stage of disease is small (for
example, the diameter of the spots of the early stage of wheat yellow rust is only 1∼2 mm),
it is difficult to distinguish the early stages of crop disease using satellite remote sensing
image data (spatial resolution is larger than 1 m/pixel). The image data provided by
satellite remote sensing can only show spectral changes when the chlorophyll and carotene
in crop cells in large fields change significantly at the end of the disease, but at this stage, it
may be too late to effectively deal with the crops, resulting in irreparable loss of crop yields.
In recent years, UAV remote sensing has been used to observe crop diseases by
acquiring near-ground high-resolution images, providing a feasible method for automatic
disease observation and prediction in the field. In the articles we reviewed, the application
of UAV remote sensing in crop diseases covered a variety of crops. Many crop disease
identification methods still rely on certain vegetation indices, but some studies started to
use neural networks instead of vegetation indices to achieve high-accuracy identification
of diseased crops. Table 4 lists part of the reviewed articles on crop diseases, comparing the
drone platforms, sensors, species, disease, and methods of crop disease identification.
Abdulridha et al. [80] proposed a method for distinguishing diseases and pests that are
visually similar to tomatoes in the field based on hyper-spectral imagery collected by UAVs.
One severe disease of citrus is huanglongbing (HLB), also known as citrus green
disease. Figure 10 shows the symptoms of huanglongbing (HLB). Chang et al. [124]
investigated the differences between healthy and diseased orange trees with four vegetation
indices: NDVI [9], NDRE [125], MSAVI [13], and chlorophyll index (CI) [126]. NDRE and
CI, which are related to the red-edge band, are more capable of monitoring citrus greening
disease. In addition, the volume of CHM is also a valuable indicator to distinguish normal
and diseased plants. Deng et al. [127] proposed a multi-feature fusion citrus yellow dragon
disease recognition method based on hyper-spectral camera data, and the recognition
accuracy on the verification dataset reached 99.73%. Moriya et al. [82] examined the
spectral band that is most easily distinguished for HLB disease using hyper-spectral camera
data. The authors found that the reflected energy of the diseased strain was significantly
stronger than that of the normal strain at 460 nm.

Figure 10. Symptoms of huanglongbing (HLB), also known as citrus green disease.

Figure 11 shows symptoms of grape disease. Kerkech et al. [128] proposed a neural net-
work method for detecting esca disease plants from grape fields based on UAV RGB image
data. By using several energy difference-related vegetation indexes (ExR [129], ExG [130],
and ExGR [131]) of diseased and normal plants reflecting light at different wavebands
as key features, the authors achieved a recognition accuracy of 95.80%. Kerkech et al. [83]
proposed a neural network method for grape disease detection based on UAV remote
sensing multi-spectral data. The authors constructed a DSM model of grape fields based on
UAV remote sensing images and combined the infrared and visible spectra of image data to
detect diseased plants in grape fields. The recognition accuracy of this method was 93.72%.
Drones 2023, 7, 398 19 of 42

Figure 11. Symptoms of grape disease.

Figure 12 shows symptoms of wheat yellow rust disease. Su et al. [132] proposed a
method for identifying wheat yellow rust based on UAV remote sensing multi-spectral
data. In the experiment, the authors found that when the spatial resolution of the image
reached 1∼1.5 cm/pixel, the UAV remote sensing method could provide enough spectral
information to distinguish wheat yellow rust. The recognition accuracy of the random forest
classifier was 89.2%. The authors found that RVI [10], NDVI [9], and OSAVI [15] were three
most effective vegetation index methods to distinguish healthy wheat from wheat yellow
rust plants. Zhang et al. [133] proposed a neural network method for detection of wheat
yellow rust plants based on UAV remote sensing hyper-spectral image data. The method’s
recognition accuracy rate was 85%. Zhang et al. [134] proposed a neural network method
for detecting wheat yellow rust based on UAV remote sensing multi-spectral image data.
This method improved the network performance by adding an irregular encoding module
(IEM), irregular decoding module (IDM), and channel weighting module (CCRM) to the
U-Net [23] structure. Compared with U-Net [23], this proposed network achieved a higher
recognition accuracy on multi-spectral image data of wheat yellow rust. Huang et al. [135]
proposed a wheat helminthosporium leaf batch disease identification method based on
UAV remote sensing RGB image data.

Figure 12. Symptoms of wheat yellow rust disease.

Kharim et al. [136] proposed a method to predict the invasion degree of bacterial
leaf blight (BLB), bacterial spike light (BPB), and stem borer (SB) in rice fields based on
UAV remote sensing RGB image data. The authors named their method the IPCA-RGB
vegetation index (IPCA− RGB ) to determine the chlorophyll content in rice plant leaves,
which is positively correlated with the degree of damage of rice plant from BLB, BPB,
and SB. Stewart et al. [137] proposed a quantitative method of maize northern leaf blight
(NLB) based on UAV remote sensing RGB data. The authors conducted experiments
based on the validation data set of the MASK-RCNN [24] neural network model, and
the obtained intersection over union (IoU) was 73%. Shegoma et al. [138] established the
Fall Armywords (FAW) infection dataset based on RGB images collected by UAVs. Based
on this dataset, the authors investigated four neural networks: VGG16 and VGG19 [139],
InceptionV3 [140], and MobileNetV2 [141]. Through transfer learning, these four methods
achieved a higher accuracy on this dataset.
Drones 2023, 7, 398 20 of 42

Table 4. Papers reviewed about crop disease observation.

Year Authors UAV Type Sensor Species Disease Method Type

2020 Abdulridha et al. [80] Multi-rotor Hyper-spectral Tomato TYLC, BS, and TS 1 SVM
2020 Chang et al. [124] Multi-rotor Multi-spectral Citrus HLB 2 Vegetation index
2020 Deng et al. [127] Multi-rotor Hyper-spectral Citrus HLB 2 Neural Network
2021 Moriya et al. [82] Multi-rotor Hyper-spectral Citrus HLB 2 Spectral Feature
2018 Kerkech et al. [128] Multi-rotor RGB Grape Virus or Fungi Neural Network
2020 Kerkech et al. [83] Multi-rotor Multi-spectral Grape Virus or Fungi Neural Network
2018 Su et al. [132] Multi-rotor Multi-spectral Wheat Wheat Yellow Rust Random Forest
2019 Zhang et al. [133] Multi-rotor Hyper-spectral Wheat Wheat Yellow Rust Neural Network
2021 Zhang et al. [134] Multi-rotor Multi-spectral Wheat Wheat Yellow Rust Neural Network
2019 Huang et al. [135] Multi-rotor RGB Wheat Helminthosporium leaf batch Neural Network
2022 Kharim et al. [136] Multi-rotor RGB Rice BLB, BPB, and SB 3 Vegetation index
2019 Stewart et al. [137] Multi-rotor RGB Corn Northern Leaf Blight (NLB) Neural Network
2021 Shegoma et al. [138] Multi-rotor RGB Corn Fall Armywords (FAW) Neural Network
2020 Ye et al. [142] Multi-rotor Multi-spectral Banana Banana Wilt Vegetation index
2019 Tetila et al. [143] Multi-rotor Multi-spectral Soybean Soybean Leaf Disease Neural Network
2017 Ha et al. [144] Multi-rotor Multi-spectral Radish Radish Wilt Neural Network
1 TYLC: yellow leaf curl, BS: bacterial spot, TS: target spot. 2 HLB: huanglongbing, also known as citrus green disease. 3 BLB: bacterial leaf blight; BPB: bacterial spike light, SB:
stem borer.
Drones 2023, 7, 398 21 of 42

Ye et al. [142] studied the multi-spectral characteristics of banana wilt disease plants
in two fields in Guangxi and Hainan Province, China. They found that the chlorophyll
content of the banana leaf and body surface would decrease significantly with the develop-
ment of fusarium wilt, and the red edge frequency in the multi-spectral data was highly
sensitive to the change in chlorophyll. This method is based on several vegetation indexes
(green chlorophyll index (CIgreen ) and red-edge chlorophyll index(CIRE ) [126], NDVI [9],
NDRE [125]) of UAV aerial multi-spectral images. The authors obtained a 80% accuracy
with a binary regression classifier.Tetila et al. [143] proposed a recognition method for
soybean leaf disease. The method is based on a RGB image taken by a UAV using SLIC
segmentation method and a neural network classifier. In the experiment, image data with a
spatial resolution of 0.85 mm/pixel were acquired at a height of 2 m above the field, and the
accuracy rate was 99.04%. Ha et al. [144] proposed a neural network method for detecting
radish wilt disease based on field RGB data photographed by a UAV.

4.2.2. Soil Water Content

Lu et al. [145] proposed a method to estimate grassland soil water content based on
UAV remote sensing RGB image data. In the study, the authors verified the feasibility of
estimating soil moisture at 0∼10 cm depth using a RGB image and established a model to
estimate soil moisture through linear regression, with a correlation coefficient R2 = 86%.
Ge et al. [146] proposed a method to estimate the soil water content of agricultural land based
on hyper-spectral remote sensing image data of UAVs. By using a XQBoost [97] classifier, the
correlation coefficient R2 = 83.5% was obtained. Bertalan et al. [147] proposed a method to
estimate soil water content based on the remote sensing multi-spectral and thermal imaging
data of UAVs. In the study, the authors found that the effect of estimating soil water content
using multi-spectral data was better than that using thermal imaging data, with a correlation
coefficient R2 = 96%. Datta et al. [48] studied the relationship between hyper-spectral image
data and soil moisture content. Based on all hyper-spectral band (AHSB) data, the authors
used support vector regression (SVR) [148], which achieved a result correlation coefficient
R2 = 95.43%, RMSE = 0.8. The authors also found that using the visible and infrared bands
(454∼742 nm) achieved the approximate same estimation results as using all hyper-spectral
band (AHSB) data. Zhang et al. [149] studied the estimation of soil moisture content in corn
fields based on multi-spectral, RGB, and thermal infrared (TIR) image data.

4.2.3. Weed Detection

Zhang et al. [150] proposed a method to observe and estimate the distribution of the
harmful plant Oxytropis ochrocephala Bunge based on UAV remote sensing image. Lan
et al. [151] proposed two neural network models, MobileNetv2-UNet and FFB-BiseNetV2,
to monitor weeds in farmland. MobileNetv2-UNet is a small model neural network, which
can be run on edge computer hardware. FFB-BiSeNetV2 is a larger parameter neural
network method that is used to identify weed targets with high accuracy at the ground end.

4.2.4. Crop Monitoring

In recent years, the use of UAV remote sensing has enabled crop monitoring and
yield estimation with an increased precision and efficiency. Several studies have explored
different approaches to obtaining valuable information from UAV remote sensing data,
including crop distribution, fruit density and distribution, biomass estimation, and crop
water content.
Lu et al. [152] proposed two methods for creating rice height models. Digital surface
point cloud (DSPC) is a model that includes the surface ground, vegetation canopy, and
water surface, while digital terrain point cloud (DTPC) represents surface cloud points
without any vegetation. The first method subtracts the DTPC from the DSPC to obtain
the height of vegetation on the ground. The second method measures vegetation height
using depth sensors on a DSPC. The height difference between the water surface reference
layer and the crop canopy is calculated to obtain the crop height. Through experiments,
Drones 2023, 7, 398 22 of 42

the authors verified that both methods can provide reliable crop height estimates, and the
second method is better. Wei et al. [153] verified the high accuracy of an existing neural
network model for identifying rice panicles from UAV remote sensing RGB images. In
the experiment, the authors used the YOLOv4 [111] neural network model to detect rice
panicles, with an average accuracy of 98.84%. The accuracy of identifying rice ears was
95.42% in the rice field without any pathogen infection, which was slightly infected. The
accuracy of data set recognition was 98.84%. The accuracy of identification of moderately
infected rice was 94.35%. The accuracy of identification of severely infected rice data was
93.36%. Cao et al. [154] compared the differences between RGB and multi-spectral data
obtained based on UAV aerial photography on the green phenotype of wheat. In compara-
tive experiments, the multi-spectral index, including red edge and infrared information,
was more effective than the color index with only visible light for wheat phenotypic clas-
sification. Zhao et al. [155] proposed a wheat spike detection method based on a UAV
remote sensing RGB image number neural network method. The authors’ method was
based on the YOLOv5 [156] network model and improved the weight post-processing
part of the network’s conjecture results. During experimental verification, the proposed
method achieved a significant improvement on the recognition accuracy of the original
YOLOv5 [156] structure network, with an overall accuracy of 94.1%, and the processing
speed was consistent with YOLOv5 [156] under the condition of processing different resolu-
tion images. Wang et al. [157] proposed a method for estimating the chlorophyll content of
winter wheat based on remote sensing multi-spectral image data from UAVs. Based on the
remote sensing image data of winter wheat taken by a multi-spectral camera, the authors
tested 26 different remote sensing indexes combined with four estimation methods, such as
random forest regression. After experimental comparison, an RF-SVR-sigmoid model, a
combination of RF variable selection and an SVR algorithm based on the sigmoid kernel,
achieved a good accuracy when estimating wheat soil plant analysis development canopy
data. Nazeri et al. [158] proposed a neural network detection method for outliers of 3D
point cloud signals based on UAV LIDAR data. Based on f-score, this method successfully
removed different levels of outliers in a crop LIDAR 3D point cloud. This method was
proven effective in experiments on sorghum and maize plants. Chen et al. [159] proposed
a method to detect agricultural crop rows in UAV images. The accuracy rate of detecting
corn planting lines with UAV remote sensing RGB image was higher than 95.45%. When
identifying wheat planting lines, the recall rate was more than 96%, and the accuracy rate
was more than 89%. Wang et al. [160] proposed a method to detect the fluorescence ratio
index (FRI) and fluorescence difference index (FDI) of rice flowering numbers per unit area
with hyper-spectral image data. Through this index system, rice yield was estimated by
taking multi-spectral images to detect the early stage of rice flowering. Traore et al. [161]
proposed a neural network method based on multi-spectral remote sensing data to detect
equivalent water thickness (EWT). Ndlovu et al. [162] proposed a random forest regression
method based on multi-spectral remote sensing image data to estimate corn water content.
Padua et al. [163] proposed an OBIA method for the classification of vineyard objects,
which included four categories: soil, shadow, other vegetation, and grape plants. In the
classification stage of the method, the author experimented with three different classifiers
(support vector machine (SVM), random forest (RF), and artificial neural network (ANN)),
and the ANN achieved the highest accuracy.

4.3. Power Lines, Towers, and Accessories

In recent years, the detection of power lines, towers, and accessories has become an
increasing industry application. A power line tower system is exposed in the field or wild
areas. After a period of time, some cables and insulators made of ceramic or glass can
become damaged through natural causes. Therefore, the power line tower must be inspected
regularly, to replace the damaged parts. Before the rise of UAV remote sensing, the standard
method of patrolling and maintaining power line towers depended on regularly arranging
workers to climb the towers. However, due to the height of power towers, and many of them
Drones 2023, 7, 398 23 of 42

being located in remote mountains and highlands, it is labor-intensive and inefficient work to
check and maintain the towers one by one. The operation becomes extremely dangerous in
winter or the rainy season. UAV remote sensing has greatly reduced the difficulty of power
line and tower inspection. Table 5 lists part of the reviewed remote sensing articles related to
power line towers and compares the data research on drone platforms, sensors, observation
objects, and method purpose. Figure 13 shows the organization of the applications for power
lines, towers, and accessories.

Figure 13. UAV remote sensing research of power lines and accessories.

The inspection objects in the research papers we reviewed mainly included transmis-
sion lines, towers, insulators, and other accessories.

4.3.1. Detection of Power Lines

As shown in Figure 14, power lines have unique image texture features in remote
sensing images. The power towers located in the field also have unique material, height,
and composition characteristics compared to the surrounding environment.

Figure 14. Power lines and tower.

Zhang et al. [164] proposed a power line detection method based on UAV remote sensing
RGB image data. In the study, the authors also disclosed two datasets: a power line dataset
of urban scenes, and a power line dataset of mountain scenes, which were UAV remote
Drones 2023, 7, 398 24 of 42

sensing power line datasets in urban and outdoor environments. The method proposed by
the authors is based on an improvement of the VGG16 neural network model [139], and the
f-score value was 91.4% on the proposed dataset. Pastucha et al. [165] proposed a power line
detection method based on UAV remote sensing RGB image data. The method was validated
on two open-source datasets and reached accuracies of 98.96% and 92.16%, respectively.
Chen et al. [54] proposed a method of detecting transmission lines based on UAV
remote sensing LIDAR data. The LIDAR equipment carried by the UAV obtained a laser
spot density of 35 points/m2 due to low-altitude flights, which is better than the detection
density that can be obtained by manned aircraft. With this method, the detection accuracy
rate was 96.5%, the recall rate was 94.8%, and f-score was 95.6%. Tan et al. [166] proposed
a method to detect transmission lines based on UAV remote sensing LIDAR data. The
authors established four datasets to verify the performance of the method. The LIDAR
point sampling density ranged from 215 points/m2 to 685 points/m2 , which is significantly
higher than the previous datasets. The accuracy of the method was higher than 97.6% on
four different datasets, and the recall rate was higher than 98.8%.
Zhang et al. [167] proposed a power line automatic measurement method based on
epipolar constraints. First, the authors acquired the spatial position of the power lines.
Second, semi-patch matching based on an epipolar constraint dense matching method was
applied to automatically extract dense point clouds within the power line corridor. Then,
obstacles were automatically detected by calculating the spatial distance between a power
line and the ground, which was represented by a 3D point cloud. This method generated 3D
point cloud data based on UAV RGB data through the SfM-MVS method, with a detection
accuracy of 93.2%, which is equivalent to manual measurement. The above results show that
this method could replace manual measurement. Zhou et al. [168] proposed an automatic
power line inspection system based on binocular vision. This method used binocular
cameras to identify the direction of the power line and then calculate its own course.

4.3.2. Detection of Power Towers

Zhang et al. [55] proposed a power tower detection method based on near-earth
remote sensing LIDAR point cloud data. The authors used an unmanned helicopter as
the remote sensing platform, flying at heights of 180 m and 220 m at different moments,
and the laser spot detection density was 30.7 points/m2 and 42.2 points/m2 respectively,
to detect the power line area. The accuracy rate of the results obtained was 96.5%, the
recall rate was 96%, and the f-score was 96.4%. Ortega et al. [169] proposed a method to
detect power tower and lines based on helicopter LIDAR data. The authors used Riegl
VUX-1LR LIDAR equipment. The helicopter flight height was 300 m, the LIDAR sampling
density was 13 points/m2 , the accuracy rate of tower recognition was 90.77%, and the
recall rate was 94.90%. This method had an accuracy of 99.44% and a recall rate of 99.58%
for the detection of power lines. Lu et al. [90] proposed a method to detect the tilt state
of transmission towers based on UAV LIDAR data. The authors inspected the tilt degree
of the body and head of the power tower in operation, to detect the safety status of the
power tower.

4.3.3. Detection of Insulators and Other Accessories

Insulators and springs are vulnerable components in power lines and are key objects
in power line inspections, requiring identification of their condition.
Figure 15 shows insulators on power lines. Zhao et al. [170] proposed a power
line tower insulator detection method based on RGB image data. Based on the R-CNN
neural network framework [171], the accuracy of detecting insulators with different lengths
reached 81.8%. Ma et al. [172] proposed a circuit tower insulator detection method based on
UAV binocular vision. The image depth information of the insulator was calculated using
the remote sensing image from a dual-view camera to assist in the accurate identification of
the insulators. The accuracy of the authors’ method was higher than 91.9%. Liu et al. [173]
proposed a neural network detection method for power tower insulators based on UAV
Drones 2023, 7, 398 25 of 42

RGB images. The authors collected and sorted the data set “CCIN detection” of power line
tower insulators and trained MTI-YOLO, a neural network model that can be deployed on
edge computing equipment, to detect insulators in UAV images. Prates et al. [174] proposed
an insulator defect identification method for power line towers based on UAV RGB image
data. The authors constructed an insulator image dataset with more than 2500 pictures.
The accuracy of the method proposed by the authors was 92% in identifying insulator types
and 85% in detecting defective insulators. Wang et al. [175] proposed a neural network
method to detect defects of circuit insulators. The detection accuracy was 98.38%, and
12.8 pictures could be detected per second. Wen et al. [176] proposed a neural network
detection method for the insulator defects of power line towers based on UAV RGB images.
The authors used a two-level neural network to identify the defects in the insulators of a
power line tower. The first-level R-CNN [171] network detects the position of the insulator
from the remote sensing image, and the second-level coder–decoder network accurately
identifies the defect position. In the authors’ experiment, the recognition accuracy was
88.7%. Chen et al. [177] proposed a method to generate wire insulator image data based on
UAV RGB images. Aiming at the characteristics of a low spatial resolution and less training
data of insulator image data of power line towers obtained from UAV RGB images, the
authors proposed a method to generate high-resolution and realistic insulator detection
images with a conditional GAN [178] for expanding training data. Liu et al. [179] proposed
a method for autonomous inspection of power line towers using UAVs.

Table 5. Papers reviewed on power lines, towers, and accessories.

Year Authors UAV Type Sensor Object Propose

2019 Zhang et al. [164] Multi-rotor RGB Power Lines Detection
2020 Pastucha et al. [165] Multi-rotor RGB Power Lines Detection
2018 Chen et al. [54] Multi-rotor LiDAR Power Lines Detection
2021 Tan et al. [166] Multi-rotor LiDAR Power Lines Detection
2017 Zhang et al. [167] Multi-rotor RGB Power Lines Auto-measurement
2022 Zhou et al. [168] Multi-rotor Binocular Vision Power Lines Auto-inspection
2019 Zhang et al. [55] Unmanned Helicopter LIDAR Power Tower Detection
2019 Ortega et al. [169] Helicopter LIDAR Power Tower & Lines Detection
2022 Lu et al. [90] Multi-rotor LIDAR Power Tower Tilt State
2019 Zhao et al. [170] Multi-Rotor RGB Insulators Detection
2021 Ma et al. [172] Multi-Rotor Binocular Vision Insulators Detection
2021 Liu et al. [173] Multi-Rotor RGB Insulators Detection
2019 Prates et al. [174] Multi-Rotor RGB Insulators Detection
2020 Wang et al. [175] Multi-Rotor RGB Insulators Detection
2021 Wen et al. [176] Multi-Rotor RGB Insulators Detection
2021 Bao et al. [180] Multi-rotor RGB Shock Absorber Detection
2022 Bao et al. [181] Multi-rotor RGB Shock Absorber Detection
Drones 2023, 7, 398 26 of 42

Figure 15. Insulators on power lines.

Figure 16 shows a shock absorber on power lines. Bao et al. [180] proposed an
improved neural network model based on the YOLOv4 [111] network to identify shock
absorbers of transmission lines on power towers. After a period of experimentation,
Bao et al. [181] put forward an improved neural network to identify transmission line
shock absorbers based on the YOLOv5 [156] network model and using the dataset they had
collected before.

Figure 16. Shock absorbers on power lines.

4.4. Buildings, Artificial Facilities, Natural Environments, and Others

UAV remote sensing has been increasingly applied in detection and information
extraction tasks for buildings, artificial targets, and other objects, due to its multi-altitude
and multi-angle observation capabilities and the ability to carry 3D sensors that can provide
elevation information of observation targets. Figure 17 shows the organization of the
applications in buildings, artificial facilities, and natural environments.
Drones 2023, 7, 398 27 of 42

Figure 17. UAV remote sensing research on artificial facilities and natural environments.

4.4.1. Buildings and Other Artificial Facilities

In recent years, a trend in the research on remote sensing of buildings has been to use
high-resolution image data to establish elevation information and the SfM-MVS method
for further processing. These research works focused on estimating the status of buildings
after disasters, identifying disaster zones, change detection, identifying building types,
and others. Nex et al. [182] proposed a method for detecting damaged buildings and
their ground coverage areas after disasters based on UAV RGB images. The authors used
multiple overlapping images to generate a sparse point cloud and using the SfM-MVS
method to detect damaged buildings. Yeom et al. [183] proposed a method for estimating
the degree of building damage caused by hurricanes based on UAV remote sensing RGB
images. Li et al. [184] proposed a method for 3D change detection of buildings based on
UAV RGB images. The authors used the SfM-MVS method to generate a digital elevation
model (DEM) and detect changes in buildings. Wu et al. [185] proposed a method for
detecting building types in built-up areas based on RGB images. The authors classified
buildings into four categories based on their number of floors: one floor, two floors, three to
six floors, and above six floors. The authors established a DEM model using the SfM-MVS
method and classified buildings based on elevation data.
Detecting buildings and their attributes based on UAV image data is also a practical
research direction. Zheng et al. [186] proposed a neural network method for detecting
buildings in UAV RGB images. Li et al. [187] proposed a neural network method for
detecting damaged buildings in RGB images. Based on the SSD [188] network model, the
authors achieved a high accuracy in detecting damaged buildings by adding a CBAM mod-
ule [189]. Xu et al. [190] proposed an encoding and decoding semantic segmentation neural
network based on a channel space enhanced attention mechanism for accurate detection
and segmentation of blue rooftop buildings from drone remote sensing RGB images.
As UAVs can observe buildings and artificial facilities from a predefined height and
angle, there have been applications focused on 3D reconstruction of buildings in recent
years. For this aim, a facade image can be used for 3D reconstruction. He et al. [191]
proposed a method to accurately map the exterior texture images of real buildings onto
their virtual models. The authors used multi-angle images of the exterior of the building,
and through geometric transformation, pasted the transformed texture image of the exterior
wall of the building onto the virtual model of the building. Zhu et al. [67] proposed a
method based on a neural network to match remote sensing images of the same location
taken from different angles and at different resolutions. By segmenting and combining the
target image and calculating the loss function separately for the local area of the image, the
authors realized a cross-view spatial geographic location method.
The 3D reconstruction technology based on UAV remote sensing has been applied to
archaeology and ancient building mapping. Alshawabkeh et al. [56] proposed an image and
3D point cloud fusion algorithm for reconstructing complex architectural scenes. Based on a
LIDAR point cloud and image data collected by UAVs, the authors were able to reconstruct
an ancient village in three dimensions. Laugier et al. [192] proposed an archaeological
Drones 2023, 7, 398 28 of 42

remote sensing method based on multiple data sources. Based on various data sources,
such as aerial remote sensing images, satellite remote sensing images, and drone remote
sensing images, the authors conducted a remote sensing survey of pre-modern land use
information in the Irwan Valley of Iraq. The targets included canals, karez, orbits, and
field systems.
UAV remote sensing is also used to detect larger vehicles, such as cars and ships. Am-
mour et al. [193] proposed a neural network method for detecting cars in UAV RGB images.
The authors divided the images into multiple blocks using the mean shift method, extracted
regional image features using a VGG network, and then identified whether a certain area
contained car targets using a linear SVM classifier. Li et al. [194] proposed a method for es-
timating the ground speed of multi-vehicles from UAV video. Zhang et al. [195] proposed
a neural network method CFF-SDN for detecting ship targets based on UAV RGB images.
The network model proposed by the authors has three branches that respectively examine
large, medium, and small targets, which can adapt to ship detection tasks of different sizes
in different sea areas and can detect overlapping targets.

4.4.2. Natural Environments and Others

UAV remote sensing has been widely used for environmental research. In the papers
we reviewed, the contents were varied. The natural environment investigations included
predicting the probability of landslides, detecting rockfalls in mountainous areas, observing
glacier changes, and observing rivers and their environment near sea outlets.
Micieli et al. [87] proposed a method of observing the water resources of rivers with
UAV thermal cameras. Compared with satellite remote sensing, the data collected by
drone sensors can have a higher temporal resolution. Lubczonek et al. [88] developed a
bathymetric method for a coastline’s critical area, combining underwater acoustic data and
UAV remote sensing image data. Lubczonek et al. [196] proposed a method for measuring
the topographical surface of shoals and coastal areas. Ioli et al. [197] used UAVs to monitor
glaciers for seven years. The authors found that using UAVs allowed effectively monitoring
the dynamics of glaciers, which is currently impossible with other observation platforms,
including remote sensing satellites. Nardin et al. [198] proposed a method for investigating
the seasonal changes of salina vegetation based on UAV images and ground survey data.
UAV remote sensing has been applied in the investigation of wild animals, with meth-
ods for detecting animals such as wild boars and marine animal activities. Kim et al. [199]
proposed a method to monitor wild boars in plains, mountainous areas, and forested
areas with thin tree canopies using infrared cameras carried by UAVs. Ranvcic et al. [200]
used the YOLOv4 neural network to identify and count deer in low-altitude forest scenes,
achieving an accuracy of 70.45% and a recall rate of 75%. Christie et al. [89] proposed
an effective method to measure the morphology of small dolphin species based on UAV
RGB images.

5. Discussion
According to the different application types of drone remote sensing, comparing
some of the reviewed papers can reveal the commonalities and differences in tasks, data
collection, data processing, and other stages of these studies.
In recent years, there have been several reviews [29–31] on UAV remote sensing.
Osco et al. [32] focused on the deep learning methods applied in UAV remote sensing.
Aasen et al. [33] focused on the data processing of hyper-spectral UAV remote sensing.
Guimaraes et al. [34] and Torresan et al. [35] focused on the application of UAV remote
sensing in forestry. Maes et al. [36] and Tsouros et al. [37] focused applications in precision
agriculture. Jafarbiglu et al. [40] reviewed UAV remote sensing of nut crops.
In this review, we mainly reviewed research papers published in the past three years
on all application fields and data processing methods for UAV remote sensing. Our goal
was to grasp the current status of the hardware, software, and data processing methods
Drones 2023, 7, 398 29 of 42

used in UAV remote sensing research, as well as the main application directions, in order
to analyze the future development direction of this research field.

5.1. Forestry Remote Sensing

Judging the differences between UAV remote sensing and satellite remote sensing in
forestry remote sensing, UAVs can set the flight height, carry LIDAR sensors for remote
sensing, and have advantages over satellite remote sensing in forest parameter measure-
ment and estimation. In disease monitoring, high-resolution images from UAV remote
sensing can produce better accuracy results.
In terms of forest parameter estimation, Ganz et al. [91] used RGB images and
Krause et al. [51] used LIDAR data from UAV remote sensing to measure tree height, and
the RMSEs obtained were 0.479 m and 0.36 m, respectively; However, Ge et al. [201] mea-
sured forest tree height based on satellite SAR data, and the RMSE obtained was as high as
25%. The accuracy of drone remote sensing in measuring an important parameter of forest
remote sensing, tree height, is significantly higher than that of satellite remote sensing.
In terms of forestry disease monitoring, taking the monitoring of pine blight as an
example, Ren et al. [84] proposed a method based on UAV remote sensing RGB images
with an accuracy of 79.8%; Li et al. [110] proposed a method based on UAV remote sensing
hyper-spectral data, with an accuracy from 84% to 99.8%; However, Zhang et al. [202] used
data obtained from remote sensing satellites, and their accuracy rate for similar diseases
was only 67.7%.
Compared with satellites, UAV remote sensing methods have a higher accuracy in
forest parameter estimation because the sensors can directly measure target elevation infor-
mation. In the forest disease monitoring problem, due to the spatial resolution advantage of
UAV remote sensing image data, it is also difficult for satellite remote sensing data methods
to compete.
From the perspective of the application of UAV remote sensing in forestry remote
sensing, and through a comparison of parameters in Table 3, we can notice some differences
and connections in the observation platforms, sensors, and information processing methods:
(1) Only two articles [109,120] used fixed-wing UAVs, but in these two studies, the data
sampling range was significantly more extensive than that of the multi-rotor methods, the
flying height was higher, and RGB sensors were used; (2) Yu et el. [85,113,114] used multi-
spectral and hyper-spectral LIDAR as sensors in their three studies, all with multi-rotor
UAVs; (3) In the research papers on pine wilt, except for the article [114] using LIDAR
data and another OBIA method [112], the research papers using RGB, multi-spectral, and
hyper-spectral all used neural networks. (4) No article used the flight data of drones,
including GNSS, flight speed, and other information.
Based on the above two perspectives, we summarize the characteristics of forestry
UAV remote sensing: (1) Due to the higher spatial resolution of satellite remote sensing
data, UAV remote sensing can achieve higher accuracy in remote sensing monitoring of
forestry diseases; (2) UAV remote sensing method has the characteristics of carrying LIDAR
and flying according to a set altitude, so it can achieve better accuracy in measuring forest
parameters than satellite remote sensing; (3) Fixed UAVs can be used as vehicles for large-
area forest remote sensing. However, this vehicle also needs to be set to fly at a higher
spatial position, and the spatial resolution of the acquired image data will be relatively low;
(4) In current research, multi-rotor UAVs are often equipped with multi-spectral cameras,
hyper-spectral cameras, and LIDARs; (5) For the data processing methods of collected RGB,
multi-spectral, and hyper-spectral sensors, the neural network method is the mainstream
method in current research; however, LIDAR data have special elevation information, and
the processing method is still relatively cumbersome; (6) Current research papers lack the
processing and utilization of UAV flight data, such as GNSS, azimuth, flight speed, and
other information.
Drones 2023, 7, 398 30 of 42

5.2. Precision Agriculture

In precision agriculture, satellite plant remote sensing data are considered insufficient
to support an accurate assessment of crop growth [203]. One of the significant achieve-
ments of UAV remote sensing in recent years is crop disease detection and classification.
Compared to the 30 m/pixel resolution image of the Landsat-5 satellite and the 10 m/pixel
resolution image of the Sentinel-2 satellite, UAV remote sensing can produce image data
with a spatial resolution up to decimeters/pixel or even centimeters/pixel.
Take the studies on wheat yellow rust as an example. The study of [204] based on
satellite remote sensing data could only verify the effectiveness of the vegetation indexes
on a 10 × 10 m field; The study of [132] based on UAV remote sensing multi-spectral images
with a spatial resolution of 1–1.5 cm/pixel could precisely identify the most relevant
changes in different spectrums of the disease. The accuracy of wheat yellow rust image
sample classification and recognition in a 1 × 1 m farmland area was 89.2%, significantly
better than the method based on satellite image data. Bohnenkamp et al. [205] studied
the relationship between the UAV-based hyper-spectral method and ground handheld
instruments. From the perspective of spectral features, the observation and recognition of
the crop canopy based on UAV remote sensing already has a similar effectiveness compared
to handheld ground methods.
Comparing the data in Table 4, we can determine some rules: (1) Articles applied to
crop disease monitoring generally used multi-rotor drones; (2) Most papers used multi-
spectral cameras, followed by RGB and hyper-spectral cameras. LIDAR was not used;
(3) Most studies used neural networks as the detection method.
We can make a summary of remote sensing for precision agriculture based on these
points: (1) The high-spatial-resolutiof images that UAV remote sensing can provide can
bring higher a recognition accuracy in monitoring and identifying crop diseases than satel-
lite remote sensing data with a lower spatial resolution; (2) Multi-spectral image data are
the most extensively studied data in agricultural disease remote sensing. However, current
unmanned aerial vehicle multi-spectral remote sensing technology still lacks the ability to
identify early symptoms of fungal infections in crop leaves. With the development of higher
resolution image sensors and data fusion technology, obtaining early crop disease infection
information from drone remote sensing images will become possible. (3) The current
research has limited demand for large-scale, low-spatial-resolution data for crop disease
monitoring. The main research focused on high-spatial resolution and small ground area
remote sensing. Therefore, multi-rotor UAVs meet the needs of this application; (4) Most of
these studies were based on using neural network methods as detectors or classifiers.

5.3. Artificial Facilities

For the remote sensing of artificial facilities, the RGB and LIDAR sensors carried by
UAVs for remote sensing can establish the elevation information of the target through ALS
or SfM-MVS methods, which is difficult to achieve based on satellite remote sensing data.
For particular targets, such as power lines, with a diameter of only a few centimeters,
UAV remote sensing has shown technical advantages. Comparing Table 5, we can see some
rules: (1) For the remote sensing of power line towers, most remote sensing observation
platforms are the multi-rotor type. When LIDAR is used as a sensor, a helicopter can also
be used as a platform; (2) When the observation object is a power line, a RGB camera or
LIDAR can be selected as the sensor. However, when RGB is used as the sensor, the power
line detection is solved as a semantic segmentation method, using a neural network; but
when LIDAR is used as the sensor, the power line detection is based on three-dimensional
point cloud data, and the recognition method is more cumbersome than a neural network;
(3) The sensors are all LIDAR when the detection object is a power tower. The experimental
results show that using LIDAR data to detect power towers can provide a high accuracy
rate; (4) When the detection objects are insulators and springs, the data used in studies are
all RGB images, and the recognition method is the neural network method.
Drones 2023, 7, 398 31 of 42

5.4. Further Research Topics

Regarding the different platforms among the many studies reviewed so far, the multi-
rotor UAV was the most adopted flying observation platform. Among the reviewed
articles, these were equipped with all types of sensor. When large-scale observations are
required in environments such as forestry remote sensing, fixed-wing UAVs need to be
used as platforms.
From the perspective of data types, in precision agriculture, the most important data
source is multi-spectral imagery. Since the current research has not yet used the image
texture features of crop diseases, there is still space for improvement in crop disease
detection using UAV remote sensing. Many fungal crop diseases can cause spots on plant
leaves, and these spots’ morphology and spatial distribution vary by fungal type. In the
early and middle stages of wheat powdery mildew, spots with a diameter of only 1–2 mm
appear on the leaves of wheat plants. In the papers we reviewed, most research data
were sampled at an altitude of 100–120 m, and the spatial resolution of the multi-spectral
images was 4–8 cm/pixel. Therefore, the speckled features caused by fungi were not
visible in these images. The later stage of the disease, when a large amount of leaf cells are
affected by the fungus and photosynthesis decreases a lot, results in apparent changes in the
reflectance spectrum, which can determine whether the farmland wheat is infected or not.
The current limitations are also opportunities for future research. With the improvement
of spatial resolution of image sensors and multi-band image registration methods, drones
equipped with higher spatial resolution multi-spectral cameras can perform close-range
remote sensing of crops. Soon, researchers will be able to obtain multi-spectral image
data with a spatial resolution of millimeters per pixel. At that time, the characteristics of
fungi and other diseases on crop leaves will be observable not only from the spectrum
but also from the texture features of multi-spectral images. Currently, neural network
methods that have been extensively studied can recognize the textural features of images
with high accuracy and recall. In summary, from the perspective of data and identification
methods, the current technological development trends are creating a robust foundation
for accurately identifying crop diseases using UAV remote sensing in the future.
RGB and LIDAR are two important data sources for the remote sensing of buildings
and artificial objects. With the improvement of the resolution of image sensors, progress
can be made in observing the position, speed, and activity of smaller artificial objects.
Li et al. [194] proposed a method for estimating the ground speed of multi-vehicles from
UAV video. Saeed et al. [206] proposed a small neural network that can run on the NVIDIA
Jetson nano embedded platform and which can be placed on a UAV, aiming to directly
detect common ground targets, including pedestrians, cars, motorcycles, direct driving,
etc. With the development of imaging camera resolution, exposure speed, and embedded
computing platforms equipped with drones, it will be possible to detect more diverse and
smaller artificial targets from UAV remote sensing image data in the future.
From the perspective of method migration, there is existing work [90] using UAVs
equipped with LIDAR sensors to realize the measurement of the tile state of power line
towers. Such methods can be widely transferred to measure other artificial structures, such
as bridges, high-rise buildings, etc. At present, in a work of [164,165] on automatic detection
of power lines by drones, since the sensor used is an RGB camera and the detection method
is a neural network, this also has the prospect of being easily migrated to scenarios such as
railway tracks and road surfaces.
Regarding data processing methods, neural networks are mostly used as detectors and
classifiers in the current research based on RGB, multi-spectral, and hyper-spectral image
data. The neural network needs to annotate the image data and can then modify a neural
network with better results based on the characteristics of the specific scene. However,
in the review of research using LIDAR data, most papers did not use neural networks to
process LIDAR data, and the current methods are still relatively complicated. Processing
UAV 3D point cloud data through neural networks may become an important feature
research direction.
Drones 2023, 7, 398 32 of 42

Fusing multi-source data is also an important development direction in UAV remote

sensing. The current research is less focused on applying multi-source and multi-type UAV
remote sensing data, such as LiDA and RGB fusion data, LiDAR and multi-spectral fusion
data, etc. Therefore, the fusion method of different source data is also a development focus.
In UAV research papers, there is only a small amount of content that used information
such as GNSS coordinates and speed during UAV flight and no research papers based on
drone video data. Data of these two aspects may become a new research hot spot.
In addition to data and processing methods, UAVs can make repeated observations of
the same ground target area at the same height and angle in the air, because they can fly
according to preset routes and take remote sensing shots at set crossing points. This feature
is suitable for research on change detection, but there is a lack of corresponding research
and applications at present.

6. Conclusions
Through this review of UAV-related papers in recent years, the authors verified that
UAV technology has been widely used in remote sensing applications such as precision
agriculture, forestry, power transmission lines, buildings, artificial objects, and natural
environments, and has shown its unique advantages. Compared with satellite remote
sensing, UAV remote sensing can provide higher resolution image data, which makes
the accuracy of crop type identification, agricultural plant disease monitoring, and crop
information extraction significantly better than when using satellite remote sensing data.
UAV LIDAR data can produce accurate elevation information for power transmission
lines, buildings, and artificial objects, which provides better results when detecting and
identifying the attributes of these targets and demonstrates that UAV remote sensing can
be used in accurate ground object identification and detection.
There are still many advantages and characteristics of UAV technology that have not
been applied in remote sensing. Considering the classification of sensors that drones can
carry, optical image data have been studied the most. With the improvement of spatial
resolution of these data, more detailed information about large targets could be extracted,
such as fungal infections on crop surfaces, or information such as the position and speed
of smaller targets. In terms of 3D stereoscopic data, multi-view stereoscopic imaging has
had more research and applications compared to LIDAR data, due to low equipment
requirements, low costs, and simple initial data processing. However, in remote sensing
tasks for buildings, bridges, iron towers, and other targets, research based on LIDAR data
will continue to be the main research object, due to its outstanding accuracy advantages.
We can find other research opportunities if we look at the current lack of usage and
processing of certain types of data from drones. The flight data of drones, such as GNSS,
flight altitude, speed, and gyroscope data during the flight, were rarely used in the research
we reviewed. The main reason for this is that the current mainstream UAV sensors lack a
data channel to link with the drone’s flight controller, so the flight controller’s data cannot
be synchronously saved to the UAV sensors.
The GNSS information of drones is crucial for accurately measuring the coordinates of
ground targets from an aerial perspective. Due to the fact that drones can achieve accurate
positioning with a horizontal error of less than 1 cm through RTK, the absolute GNSS
coordinates of ground targets can be obtained not only from the relative position of ground
targets and the drone but also from the GNSS coordinates of the drone itself, and the error
mainly depends on the relative position error measured from image and video data with
the drone.
The flight altitude of drones plays a crucial role in determining the spatial resolution
of the image sensors carried and for measuring the elevation information of ground targets.
However, in the papers we reviewed, most drones flew at a fixed altitude when collecting
data. This flight method is suitable for observing targets on flat ground. For targets that
require elevation information, rich multi-view images can be established by remote sensing
Drones 2023, 7, 398 33 of 42

the targets at different altitudes using drones, and three-dimensional information of the
targets can be reconstructed through the SfM-MVS method.
In the currently reviewed drone remote sensing articles, neither image nor video data
were synchronized with gyroscope data. However, in recent years, in the newly published
articles on SLAM, the use of high-precision gyroscopes has made relatively considerable
progress in accuracy in 3D imaging and 3D reconstruction. A drone flight controller’s
gyroscope system has advanced sensors and complete physical damping and sensor noise
elimination methods. Therefore, it could be possible that some indoor SLAM methods
could be migrated to drone platforms, to utilize the gyroscope data.
The above is a prediction of the future development direction of drone remote sensing
from different data sources and processing perspectives. In addition, drones are excellent
data acquisition and observation platforms for performing change detection tasks, due
to their ability to program flight routes and remote sensing positions and observe the
determined flight routes multiple times. Using drones, not only can observation targets be
observed multiple times from the same angle, but also unrestricted temporal resolutions
can be achieved. Therefore, we believe that change detection based on drones should
experience significant development in the next few years.
Author Contributions: Conceptualization, L.Z.; Literature view, Z.Z. and L.Z.; writing, Z.Z. and L.Z.;
editing, Z.Z. All authors have read and agreed to the published version of the manuscript.
Funding: This work was funded by the Laboratory of Lingnan Modern Agriculture Project (Grant
No. NZ2021038).
Data Availability Statement: Data available on request from the authors.
Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations

The following abbreviations are used in this manuscript:

UAV unmanned aerial vehicle

OBIA object-oriented analysis methods
TM thematic mapper
MSS multi-spectral scanner system
VTOL vertical take-off and landing
NIR near-infrared
LIDAR laser imaging, detection, and ranging
IMU inertial measurement units
GNSS global navigation satellite systems
DEM digital elevation model
DTM digital terrain model
DSM digital surface model
SfM-MVS structure from motion and multi-view stereo
ALS airbrone lisar scanning
GCP ground control point
GSD ground sample distance
CC canopy closure
HP hemispheric photography
CHM canopy-height model
SHP synthetic hemispheric photography
SLAM simultaneous localization and mapping
SLR single lens reflex
GLCM grey-level co-occurrence matrix
SPAD soil plant analysis development
LSTM long-short term memory
SOC soil organic carbon
SMC soil water content
Drones 2023, 7, 398 34 of 42

HLB huanglongbing
IEM irregular encoding module
IDM irregular decoding module
CCRM channel weighting module
AHSB all hyper-spectral bands
TIR thermal infrared
RMSE root-mean-square error
NDVI normalized difference vegetation index
NDWI normalized difference water index
EVI enhanced vegetation index
LAI leaf area index
NDRE normalized difference red edge index
SAVI soil adjusted vegetation index
MSAVI improved soil adjusted vegetation index
CI chlorophyll index
FRI fluorescence ratio index
FDI fluorescence difference index
EWT equivalent water thicknes
DSPC digital terrain point cloud
DTPC digital terrain point cloud
PLDU power line dataset of urban scene
PLDM power line dataset of mountain scene
SVM support vector machine
RF random forest
ANN artificial neural network
SVR support vector regression
PLAMEC power line automatic measurement method based on epipolar constraints
SPMEC semi patch matching based on epipolar constraints

References
1. Simonett, D.S. Future and Present Needs of Remote Sensing in Geography; Technical Report; 1966. Available online: https:
//ntrs.nasa.gov/citations/19670031579 (accessed on 23 May 2023).
2. Hudson, R.; Hudson, J.W. The military applications of remote sensing by infrared. Proc. IEEE 1975, 63, 104–128. [CrossRef]
3. Badgley, P.C. Current Status of NASA’s Natural Resources Program. Exploring Unknown. 1960; p. 226. Available online:
https://ntrs.nasa.gov/citations/19670031597 (accessed on 23 May 2023).
4. Roads, B.O.P. Remote Sensing Applications to Highway Engineering. Public Roads 1968, 35, 28.
5. Taylor, J.I.; Stingelin, R.W. Infrared imaging for water resources studies. J. Hydraul. Div. 1969, 95, 175–190. [CrossRef]
6. Roy, D.P.; Wulder, M.A.; Loveland, T.R.; Woodcock, C.E.; Allen, R.G.; Anderson, M.C.; Helder, D.; Irons, J.R.; Johnson, D.M.;
Kennedy, R.; et al. Landsat-8: Science and product vision for terrestrial global change research. Remote Sens. Environ. 2014,
145, 154–172. [CrossRef]
7. Chevrel, M.; Courtois, M.; Weill, G. The SPOT satellite remote sensing mission. Photogramm. Eng. Remote Sens. 1981, 47, 1163–1171.
8. Dial, G.; Bowen, H.; Gerlach, F.; Grodecki, J.; Oleszczuk, R. IKONOS satellite, imagery, and products. Remote Sens. Environ. 2003,
88, 23–36. [CrossRef]
9. Rouse, J.W., Jr.; Haas, R.H.; Schell, J.; Deering, D. Monitoring the Vernal Advancement and Retrogradation (Green Wave Effect) of Natural
Vegetation; Technical Report; 1973. Available online: https://ntrs.nasa.gov/citations/19740022555 (accessed on 23 May 2023).
10. Jordan, C.F. Derivation of leaf-area index from quality of light on the forest floor. Ecology 1969, 50, 663–666. [CrossRef]
11. Huete, A.; Didan, K.; Miura, T.; Rodriguez, E.P.; Gao, X.; Ferreira, L.G. Overview of the radiometric and biophysical performance
of the MODIS vegetation indices. Remote Sens. Environ. 2002, 83, 195–213. [CrossRef]
12. Gao, B.C. NDWI—A normalized difference water index for remote sensing of vegetation liquid water from space. Remote Sens.
Environ. 1996, 58, 257–266. [CrossRef]
13. Qi, J.; Chehbouni, A.; Huete, A.R.; Kerr, Y.H.; Sorooshian, S. A modified soil adjusted vegetation index. Remote Sens. Environ.
1994, 48, 119–126. [CrossRef]
14. Huete, A.R. A soil-adjusted vegetation index (SAVI). Remote Sens. Environ. 1988, 25, 295–309. [CrossRef]
15. Rondeaux, G.; Steven, M.; Baret, F. Optimization of soil-adjusted vegetation indices. Remote Sens. Environ. 1996, 55, 95–107.
[CrossRef]
16. Blaschke, T.; Lang, S.; Lorup, E.; Strobl, J.; Zeil, P. Object-oriented image processing in an integrated GIS/remote sensing
environment and perspectives for environmental applications. Environ. Inf. Plan. Politics Public 2000, 2, 555–570.
Drones 2023, 7, 398 35 of 42

17. Blaschke, T.; Strobl, J. What’s wrong with pixels? Some recent developments interfacing remote sensing and GIS. Z. Geoinforma-
tionssysteme 2001, 12–17. Available online: https://www.researchgate.net/publication/216266284_What’s_wrong_with_pixels_
Some_recent_developments_interfacing_remote_sensing_and_GIS (accessed on 23 May 2023).
18. Schiewe, J. Segmentation of high-resolution remotely sensed data-concepts, applications and problems. Int. Arch. Photogramm.
Remote Sens. Spat. Inf. Sci. 2002, 34, 380–385.
19. Hay, G.J.; Blaschke, T.; Marceau, D.J.; Bouchard, A. A comparison of three image-object methods for the multiscale analysis of
landscape structure. ISPRS J. Photogramm. Remote Sens. 2003, 57, 327–345. [CrossRef]
20. Benz, U.C.; Hofmann, P.; Willhauck, G.; Lingenfelder, I.; Heynen, M. Multi-resolution, object-oriented fuzzy analysis of remote
sensing data for GIS-ready information. ISPRS J. Photogramm. Remote Sens. 2004, 58, 239–258. [CrossRef]
21. Blaschke, T.; Burnett, C.; Pekkarinen, A. New contextual approaches using image segmentation for objectbased classification.
In Remote Sensing Image Analysis: Including the Spatial Domain; De Meer, F., de Jong, S., Eds.; 2004. Available online: https:
//courses.washington.edu/cfr530/GIS200106012.pdf (accessed on 23 May 2023).
22. Zhan, Q.; Molenaar, M.; Tempfli, K.; Shi, W. Quality assessment for geo-spatial objects derived from remotely sensed data. Int. J.
Remote Sens. 2005, 26, 2953–2974. [CrossRef]
23. Ronneberger, O.; Fischer, P.; Brox, T. U-net: Convolutional networks for biomedical image segmentation. In Proceedings of
the Medical Image Computing and Computer-Assisted Intervention–MICCAI 2015: 18th International Conference, Munich,
Germany, 5–9 October 2015; Proceedings, Part III 18; Springer: Cham, Switzerland, 2015; pp. 234–241.
24. He, K.; Gkioxari, G.; Dollár, P.; Girshick, R. Mask r-cnn. In Proceedings of the IEEE international Conference on Computer Vision,
Venice, Italy, 22–29 October 2017; pp. 2961–2969.
25. Hu, J.; Shen, L.; Sun, G. Squeeze-and-excitation networks. In Proceedings of the IEEE Conference on Computer Vision and
Pattern Recognition, Salt Lake City, UT, USA, 18–23 June 2018; pp. 7132–7141.
26. Chen, L.C.; Papandreou, G.; Kokkinos, I.; Murphy, K.; Yuille, A.L. Deeplab: Semantic image segmentation with deep convolutional
nets, atrous convolution, and fully connected crfs. IEEE Trans. Pattern Anal. Mach. Intell. 2017, 40, 834–848. [CrossRef]
27. Chu, X.; Zheng, A.; Zhang, X.; Sun, J. Detection in crowded scenes: One proposal, multiple predictions. In Proceedings of the
IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle, WA, USA, 13–19 June 2020; pp. 12214–12223.
28. Blaschke, T. Object based image analysis for remote sensing. ISPRS J. Photogramm. Remote Sens. 2010, 65, 2–16. [CrossRef]
29. Yao, H.; Qin, R.; Chen, X. Unmanned aerial vehicle for remote sensing applications—A review. Remote Sens. 2019, 11, 1443.
[CrossRef]
30. Colomina, I.; Molina, P. Unmanned aerial systems for photogrammetry and remote sensing: A review. ISPRS J. Photogramm.
Remote Sens. 2014, 92, 79–97. [CrossRef]
31. Alvarez-Vanhard, E.; Corpetti, T.; Houet, T. UAV & satellite synergies for optical remote sensing applications: A literature review.
Sci. Remote Sens. 2021, 3, 100019.
32. Osco, L.P.; Junior, J.M.; Ramos, A.P.M.; de Castro Jorge, L.A.; Fatholahi, S.N.; de Andrade Silva, J.; Matsubara, E.T.; Pistori, H.;
Gonçalves, W.N.; Li, J. A review on deep learning in UAV remote sensing. Int. J. Appl. Earth Obs. Geoinf. 2021, 102, 102456.
[CrossRef]
33. Aasen, H.; Honkavaara, E.; Lucieer, A.; Zarco-Tejada, P.J. Quantitative remote sensing at ultra-high resolution with UAV
spectroscopy: A review of sensor technology, measurement procedures, and data correction workflows. Remote Sens. 2018,
10, 1091. [CrossRef]
34. Guimarães, N.; Pádua, L.; Marques, P.; Silva, N.; Peres, E.; Sousa, J.J. Forestry remote sensing from unmanned aerial vehicles: A
review focusing on the data, processing and potentialities. Remote Sens. 2020, 12, 1046. [CrossRef]
35. Torresan, C.; Berton, A.; Carotenuto, F.; Di Gennaro, S.F.; Gioli, B.; Matese, A.; Miglietta, F.; Vagnoli, C.; Zaldei, A.; Wallace, L.
Forestry applications of UAVs in Europe: A review. Int. J. Remote Sens. 2017, 38, 2427–2447. [CrossRef]
36. Maes, W.H.; Steppe, K. Perspectives for remote sensing with unmanned aerial vehicles in precision agriculture. Trends Plant Sci.
2019, 24, 152–164. [CrossRef]
37. Tsouros, D.C.; Bibi, S.; Sarigiannidis, P.G. A review on UAV-based applications for precision agriculture. Information 2019, 10, 349.
[CrossRef]
38. Olson, D.; Anderson, J. Review on unmanned aerial vehicles, remote sensors, imagery processing, and their applications in
agriculture. Agron. J. 2021, 113, 971–992. [CrossRef]
39. Sishodia, R.P.; Ray, R.L.; Singh, S.K. Applications of remote sensing in precision agriculture: A review. Remote Sens. 2020, 12, 3136.
[CrossRef]
40. Jafarbiglu, H.; Pourreza, A. A comprehensive review of remote sensing platforms, sensors, and applications in nut crops. Comput.
Electron. Agric. 2022, 197, 106844. [CrossRef]
41. Carrivick, J.L.; Smith, M.W.; Quincey, D.J. Structure from Motion in the Geosciences; John Wiley & Sons: Hoboken, NJ, USA, 2016.
42. Liu, Y.; Zheng, X.; Ai, G.; Zhang, Y.; Zuo, Y. Generating a high-precision true digital orthophoto map based on UAV images.
ISPRS Int. J. Geo-Inf. 2018, 7, 333. [CrossRef]
43. Watson, D.J. Comparative physiological studies on the growth of field crops: I. Variation in net assimilation rate and leaf area
between species and varieties, and within and between years. Ann. Bot. 1947, 11, 41–76. [CrossRef]
Drones 2023, 7, 398 36 of 42

44. Seager, S.; Turner, E.L.; Schafer, J.; Ford, E.B. Vegetation’s red edge: A possible spectroscopic biosignature of extraterrestrial plants.
Astrobiology 2005, 5, 372–390. [CrossRef]
45. Delegido, J.; Verrelst, J.; Meza, C.; Rivera, J.; Alonso, L.; Moreno, J. A red-edge spectral index for remote sensing estimation of
green LAI over agroecosystems. Eur. J. Agron. 2013, 46, 42–52. [CrossRef]
46. Lin, S.; Li, J.; Liu, Q.; Li, L.; Zhao, J.; Yu, W. Evaluating the effectiveness of using vegetation indices based on red-edge reflectance
from Sentinel-2 to estimate gross primary productivity. Remote Sens. 2019, 11, 1303. [CrossRef]
47. Imran, H.A.; Gianelle, D.; Rocchini, D.; Dalponte, M.; Martín, M.P.; Sakowska, K.; Wohlfahrt, G.; Vescovo, L. VIS-NIR, red-edge
and NIR-shoulder based normalized vegetation indices response to co-varying leaf and Canopy structural traits in heterogeneous
grasslands. Remote Sens. 2020, 12, 2254. [CrossRef]
48. Datta, D.; Paul, M.; Murshed, M.; Teng, S.W.; Schmidtke, L. Soil Moisture, Organic Carbon, and Nitrogen Content Prediction with
Hyperspectral Data Using Regression Models. Sensors 2022, 22, 7998. [CrossRef]
49. Jackisch, R.; Madriz, Y.; Zimmermann, R.; Pirttijärvi, M.; Saartenoja, A.; Heincke, B.H.; Salmirinne, H.; Kujasalo, J.P.; Andreani, L.;
Gloaguen, R. Drone-borne hyperspectral and magnetic data integration: Otanmäki Fe-Ti-V deposit in Finland. Remote Sens. 2019,
11, 2084. [CrossRef]
50. Thiele, S.T.; Bnoulkacem, Z.; Lorenz, S.; Bordenave, A.; Menegoni, N.; Madriz, Y.; Dujoncquoy, E.; Gloaguen, R.; Kenter, J.
Mineralogical mapping with accurately corrected shortwave infrared hyperspectral data acquired obliquely from UAVs. Remote
Sens. 2021, 14, 5. [CrossRef]
51. Krause, S.; Sanders, T.G.; Mund, J.P.; Greve, K. UAV-based photogrammetric tree height measurement for intensive forest
monitoring. Remote Sens. 2019, 11, 758. [CrossRef]
52. Yu, J.W.; Yoon, Y.W.; Baek, W.K.; Jung, H.S. Forest Vertical Structure Mapping Using Two-Seasonal Optic Images and LiDAR
DSM Acquired from UAV Platform through Random Forest, XGBoost, and Support Vector Machine Approaches. Remote Sens.
2021, 13, 4282. [CrossRef]
53. Zhang, H.; Bauters, M.; Boeckx, P.; Van Oost, K. Mapping canopy heights in dense tropical forests using low-cost UAV-derived
photogrammetric point clouds and machine learning approaches. Remote Sens. 2021, 13, 3777. [CrossRef]
54. Chen, C.; Yang, B.; Song, S.; Peng, X.; Huang, R. Automatic clearance anomaly detection for transmission line corridors utilizing
UAV-Borne LIDAR data. Remote Sens. 2018, 10, 613. [CrossRef]
55. Zhang, R.; Yang, B.; Xiao, W.; Liang, F.; Liu, Y.; Wang, Z. Automatic extraction of high-voltage power transmission objects from
UAV lidar point clouds. Remote Sens. 2019, 11, 2600. [CrossRef]
56. Alshawabkeh, Y.; Baik, A.; Fallatah, A. As-Textured As-Built BIM Using Sensor Fusion, Zee Ain Historical Village as a Case Study.
Remote Sens. 2021, 13, 5135. [CrossRef]
57. Short, N.M. The Landsat Tutorial Workbook: Basics of Satellite Remote Sensing; National Aeronautics and Space Administration,
Scientific and Technical Information Branch: Washington, DC, USA, 1982; Volume 1078.
58. Schowengerdt, R.A. Soft classification and spatial-spectral mixing. In Proceedings of the International Workshop on Soft
Computing in Remote Sensing Data Analysis, Milan, Italy, 4–5 December 1995; pp. 4–5.
59. Long, J.; Shelhamer, E.; Darrell, T. Fully convolutional networks for semantic segmentation. In Proceedings of the IEEE
Conference on Computer Vision and Pattern Recognition, Boston, MA, USA, 7–12 June 2015; pp. 3431–3440.
60. Badrinarayanan, V.; Kendall, A.; Cipolla, R. Segnet: A deep convolutional encoder-decoder architecture for image segmentation.
IEEE Trans. Pattern Anal. Mach. Intell. 2017, 39, 2481–2495. [CrossRef]
61. Chen, L.C.; Zhu, Y.; Papandreou, G.; Schroff, F.; Adam, H. Encoder-decoder with atrous separable convolution for semantic image
segmentation. In Proceedings of the European Conference on Computer Vision (ECCV), Munich, Germany, 8–14 September 2018;
pp. 801–818.
62. Wang, X.; Kong, T.; Shen, C.; Jiang, Y.; Li, L. Solo: Segmenting objects by locations. In Proceedings of the Computer Vision–ECCV
2020: 16th European Conference, Glasgow, UK, 23–28 August 2020; Proceedings, Part XVIII 16; Springer: Cham, Switzerland,
2020; pp. 649–665.
63. Bolya, D.; Zhou, C.; Xiao, F.; Lee, Y.J. Yolact: Real-time instance segmentation. In Proceedings of the IEEE/CVF International
Conference on Computer Vision, Seoul, Republic of Korea, 27 October–2 November 2019; pp. 9157–9166.
64. Zhao, G.; Zhang, W.; Peng, Y.; Wu, H.; Wang, Z.; Cheng, L. PEMCNet: An Efficient Multi-Scale Point Feature Fusion Network for
3D LiDAR Point Cloud Classification. Remote Sens. 2021, 13, 4312. [CrossRef]
65. Harvey, W.; Rainwater, C.; Cothren, J. Direct Aerial Visual Geolocalization Using Deep Neural Networks. Remote Sens. 2021,
13, 4017. [CrossRef]
66. Chollet, F. Xception: Deep learning with depthwise separable convolutions. In Proceedings of the IEEE Conference on Computer
Vision and Pattern Recognition, Honolulu, HI, USA, 21–26 July 2017; pp. 1251–1258.
67. Zhuang, J.; Dai, M.; Chen, X.; Zheng, E. A Faster and More Effective Cross-View Matching Method of UAV and Satellite Images
for UAV Geolocalization. Remote Sens. 2021, 13, 3979. [CrossRef]
68. Chen, B.; Chen, Z.; Deng, L.; Duan, Y.; Zhou, J. Building change detection with RGB-D map generated from UAV images.
Neurocomputing 2016, 208, 350–364. [CrossRef]
69. Cook, K.L. An evaluation of the effectiveness of low-cost UAVs and structure from motion for geomorphic change detection.
Geomorphology 2017, 278, 195–208. [CrossRef]
Drones 2023, 7, 398 37 of 42

70. Mesquita, D.B.; dos Santos, R.F.; Macharet, D.G.; Campos, M.F.; Nascimento, E.R. Fully convolutional siamese autoencoder for
change detection in UAV aerial images. IEEE Geosci. Remote Sens. Lett. 2019, 17, 1455–1459. [CrossRef]
71. Hastaoğlu, K.Ö.; Gül, Y.; Poyraz, F.; Kara, B.C. Monitoring 3D areal displacements by a new methodology and software using
UAV photogrammetry. Int. J. Appl. Earth Obs. Geoinf. 2019, 83, 101916. [CrossRef]
72. Lucieer, A.; Jong, S.M.d.; Turner, D. Mapping landslide displacements using Structure from Motion (SfM) and image correlation
of multi-temporal UAV photography. Prog. Phys. Geogr. 2014, 38, 97–116. [CrossRef]
73. Li, M.; Cheng, D.; Yang, X.; Luo, G.; Liu, N.; Meng, C.; Peng, Q. High precision slope deformation monitoring by uav with
industrial photogrammetry. IOP Conf. Ser. Earth Environ. Sci. 2021, 636, 012015. [CrossRef]
74. Han, D.; Lee, S.B.; Song, M.; Cho, J.S. Change detection in unmanned aerial vehicle images for progress monitoring of road
construction. Buildings 2021, 11, 150. [CrossRef]
75. Huang, R.; Xu, Y.; Hoegner, L.; Stilla, U. Semantics-aided 3D change detection on construction sites using UAV-based photogram-
metric point clouds. Autom. Constr. 2022, 134, 104057. [CrossRef]
76. Sanz-Ablanedo, E.; Chandler, J.H.; Rodríguez-Pérez, J.R.; Ordóñez, C. Accuracy of unmanned aerial vehicle (UAV) and SfM
photogrammetry survey as a function of the number and location of ground control points used. Remote Sens. 2018, 10, 1606.
[CrossRef]
77. Rebelo, C.; Nascimento, J. Measurement of Soil Tillage Using UAV High-Resolution 3D Data. Remote Sens. 2021, 13, 4336.
[CrossRef]
78. Almeida, A.; Gonçalves, F.; Silva, G.; Mendonça, A.; Gonzaga, M.; Silva, J.; Souza, R.; Leite, I.; Neves, K.; Boeno, M.; et al.
Individual Tree Detection and Qualitative Inventory of a Eucalyptus sp. Stand Using UAV Photogrammetry Data. Remote Sens.
2021, 13, 3655. [CrossRef]
79. Hartwig, M.E.; Ribeiro, L.P. Gully evolution assessment from structure-from-motion, southeastern Brazil. Environ. Earth Sci.
2021, 80, 548. [CrossRef]
80. Abdulridha, J.; Ampatzidis, Y.; Qureshi, J.; Roberts, P. Laboratory and UAV-based identification and classification of tomato
yellow leaf curl, bacterial spot, and target spot diseases in tomato utilizing hyperspectral imaging and machine learning. Remote
Sens. 2020, 12, 2732. [CrossRef]
81. Ampatzidis, Y.; Partel, V. UAV-based high throughput phenotyping in citrus utilizing multispectral imaging and artificial
intelligence. Remote Sens. 2019, 11, 410. [CrossRef]
82. Moriya, É.A.S.; Imai, N.N.; Tommaselli, A.M.G.; Berveglieri, A.; Santos, G.H.; Soares, M.A.; Marino, M.; Reis, T.T. Detection
and mapping of trees infected with citrus gummosis using UAV hyperspectral data. Comput. Electron. Agric. 2021, 188, 106298.
[CrossRef]
83. Kerkech, M.; Hafiane, A.; Canals, R. VddNet: Vine disease detection network based on multispectral images and depth map.
Remote Sens. 2020, 12, 3305. [CrossRef]
84. Ren, D.; Peng, Y.; Sun, H.; Yu, M.; Yu, J.; Liu, Z. A Global Multi-Scale Channel Adaptation Network for Pine Wilt Disease Tree
Detection on UAV Imagery by Circle Sampling. Drones 2022, 6, 353. [CrossRef]
85. Yu, R.; Luo, Y.; Zhou, Q.; Zhang, X.; Wu, D.; Ren, L. Early detection of pine wilt disease using deep learning algorithms and
UAV-based multispectral imagery. For. Ecol. Manag. 2021, 497, 119493. [CrossRef]
86. Hu, G.; Yin, C.; Wan, M.; Zhang, Y.; Fang, Y. Recognition of diseased Pinus trees in UAV images using deep learning and
AdaBoost classifier. Biosyst. Eng. 2020, 194, 138–151. [CrossRef]
87. Micieli, M.; Botter, G.; Mendicino, G.; Senatore, A. UAV Thermal Images for Water Presence Detection in a Mediterranean
Headwater Catchment. Remote Sens. 2021, 14, 108. [CrossRef]
88. Lubczonek, J.; Kazimierski, W.; Zaniewicz, G.; Lacka, M. Methodology for combining data acquired by unmanned surface and
aerial vehicles to create digital bathymetric models in shallow and ultra-shallow waters. Remote Sens. 2021, 14, 105. [CrossRef]
89. Christie, A.I.; Colefax, A.P.; Cagnazzi, D. Feasibility of Using Small UAVs to Derive Morphometric Measurements of Australian
Snubfin (Orcaella heinsohni) and Humpback (Sousa sahulensis) Dolphins. Remote Sens. 2021, 14, 21. [CrossRef]
90. Lu, Z.; Gong, H.; Jin, Q.; Hu, Q.; Wang, S. A transmission tower tilt state assessment approach based on dense point cloud from
UAV-based LiDAR. Remote Sens. 2022, 14, 408. [CrossRef]
91. Ganz, S.; Käber, Y.; Adler, P. Measuring tree height with remote sensing—A comparison of photogrammetric and LiDAR data
with different field measurements. Forests 2019, 10, 694. [CrossRef]
92. Fakhri, S.A.; Latifi, H. A Consumer Grade UAV-Based Framework to Estimate Structural Attributes of Coppice and High Oak
Forest Stands in Semi-Arid Regions. Remote Sens. 2021, 13, 4367. [CrossRef]
93. Meyer, F.; Beucher, S. Morphological segmentation. J. Vis. Commun. Image Represent. 1990, 1, 21–46. [CrossRef]
94. Pu, Y.; Xu, D.; Wang, H.; An, D.; Xu, X. Extracting Canopy Closure by the CHM-Based and SHP-Based Methods with a
Hemispherical FOV from UAV-LiDAR Data in a Poplar Plantation. Remote Sens. 2021, 13, 3837. [CrossRef]
95. Mo, J.; Lan, Y.; Yang, D.; Wen, F.; Qiu, H.; Chen, X.; Deng, X. Deep Learning-Based Instance Segmentation Method of Litchi
Canopy from UAV-Acquired Images. Remote Sens. 2021, 13, 3919. [CrossRef]
96. Reder, S.; Mund, J.P.; Albert, N.; Waßermann, L.; Miranda, L. Detection of Windthrown Tree Stems on UAV-Orthomosaics Using
U-Net Convolutional Networks. Remote Sens. 2021, 14, 75. [CrossRef]
Drones 2023, 7, 398 38 of 42

97. Chen, T.; Guestrin, C. Xgboost: A scalable tree boosting system. In Proceedings of the 22nd ACM Sigkdd International Conference
on Knowledge Discovery and Data Mining, San Francisco, CA, USA, 13–17 August 2016; pp. 785–794.
98. Guo, X.; Liu, Q.; Sharma, R.P.; Chen, Q.; Ye, Q.; Tang, S.; Fu, L. Tree Recognition on the Plantation Using UAV Images with
Ultrahigh Spatial Resolution in a Complex Environment. Remote Sens. 2021, 13, 4122. [CrossRef]
99. Taylor-Zavala, R.; Ramírez-Rodríguez, O.; de Armas-Ricard, M.; Sanhueza, H.; Higueras-Fredes, F.; Mattar, C. Quantifying
Biochemical Traits over the Patagonian Sub-Antarctic Forests and Their Relation to Multispectral Vegetation Indices. Remote Sens.
2021, 13, 4232. [CrossRef]
100. Li, X.; Gao, H.; Zhang, M.; Zhang, S.; Gao, Z.; Liu, J.; Sun, S.; Hu, T.; Sun, L. Prediction of Forest Fire Spread Rate Using UAV
Images and an LSTM Model Considering the Interaction between Fire and Wind. Remote Sens. 2021, 13, 4325. [CrossRef]
101. Hochreiter, S.; Schmidhuber, J. Long short-term memory. Neural Comput. 1997, 9, 1735–1780. [CrossRef] [PubMed]
102. Hu, J.; Niu, H.; Carrasco, J.; Lennox, B.; Arvin, F. Fault-tolerant cooperative navigation of networked UAV swarms for forest fire
monitoring. Aerosp. Sci. Technol. 2022, 123, 107494. [CrossRef]
103. Namburu, A.; Selvaraj, P.; Mohan, S.; Ragavanantham, S.; Eldin, E.T. Forest Fire Identification in UAV Imagery Using X-MobileNet.
Electronics 2023, 12, 733. [CrossRef]
104. Howard, A.G.; Zhu, M.; Chen, B.; Kalenichenko, D.; Wang, W.; Weyand, T.; Andreetto, M.; Adam, H. Mobilenets: Efficient
convolutional neural networks for mobile vision applications. arXiv 2017, arXiv:1704.04861.
105. Beltrán-Marcos, D.; Suárez-Seoane, S.; Fernández-Guisuraga, J.M.; Fernández-García, V.; Marcos, E.; Calvo, L. Relevance of UAV
and sentinel-2 data fusion for estimating topsoil organic carbon after forest fire. Geoderma 2023, 430, 116290. [CrossRef]
106. Rutherford, T.; Webster, J. Distribution of pine wilt disease with respect to temperature in North America, Japan, and Europe.
Can. J. For. Res. 1987, 17, 1050–1059. [CrossRef]
107. Hunt, D. Pine wilt disease: A worldwide threat to forest ecosystems. Nematology 2009, 11, 315–316. [CrossRef]
108. Wu, B.; Liang, A.; Zhang, H.; Zhu, T.; Zou, Z.; Yang, D.; Tang, W.; Li, J.; Su, J. Application of conventional UAV-based high-
throughput object detection to the early diagnosis of pine wilt disease by deep learning. For. Ecol. Manag. 2021, 486, 118986.
[CrossRef]
109. Xia, L.; Zhang, R.; Chen, L.; Li, L.; Yi, T.; Wen, Y.; Ding, C.; Xie, C. Evaluation of Deep Learning Segmentation Models for
Detection of Pine Wilt Disease in Unmanned Aerial Vehicle Images. Remote Sens. 2021, 13, 3594. [CrossRef]
110. Li, F.; Liu, Z.; Shen, W.; Wang, Y.; Wang, Y.; Ge, C.; Sun, F.; Lan, P. A remote sensing and airborne edge-computing based detection
system for pine wilt disease. IEEE Access 2021, 9, 66346–66360. [CrossRef]
111. Bochkovskiy, A.; Wang, C.Y.; Liao, H.Y.M. Yolov4: Optimal speed and accuracy of object detection. arXiv 2020, arXiv:2004.10934.
112. Sun, Z.; Wang, Y.; Pan, L.; Xie, Y.; Zhang, B.; Liang, R.; Sun, Y. Pine wilt disease detection in high-resolution UAV images using
object-oriented classification. J. For. Res. 2022, 33, 1377–1389. [CrossRef]
113. Yu, R.; Luo, Y.; Li, H.; Yang, L.; Huang, H.; Yu, L.; Ren, L. Three-Dimensional Convolutional Neural Network Model for Early
Detection of Pine Wilt Disease Using UAV-Based Hyperspectral Images. Remote Sens. 2021, 13, 4065. [CrossRef]
114. Yu, R.; Luo, Y.; Zhou, Q.; Zhang, X.; Wu, D.; Ren, L. A machine learning algorithm to detect pine wilt disease using UAV-based
hyperspectral imagery and LiDAR data at the tree level. Int. J. Appl. Earth Obs. Geoinf. 2021, 101, 102363. [CrossRef]
115. Li, J.; Wang, X.; Zhao, H.; Hu, X.; Zhong, Y. Detecting pine wilt disease at the pixel level from high spatial and spectral resolution
UAV-borne imagery in complex forest landscapes using deep one-class classification. Int. J. Appl. Earth Obs. Geoinf. 2022,
112, 102947. [CrossRef]
116. Dash, J.P.; Watt, M.S.; Pearse, G.D.; Heaphy, M.; Dungey, H.S. Assessing very high resolution UAV imagery for monitoring forest
health during a simulated disease outbreak. ISPRS J. Photogramm. Remote Sens. 2017, 131, 1–14. [CrossRef]
117. Sandino, J.; Pegg, G.; Gonzalez, F.; Smith, G. Aerial mapping of forests affected by pathogens using UAVs, hyperspectral sensors,
and artificial intelligence. Sensors 2018, 18, 944. [CrossRef]
118. Näsi, R.; Honkavaara, E.; Blomqvist, M.; Lyytikäinen-Saarenmaa, P.; Hakala, T.; Viljanen, N.; Kantola, T.; Holopainen, M. Remote
sensing of bark beetle damage in urban forests at individual tree level using a novel hyperspectral camera from UAV and aircraft.
Urban For. Urban Green. 2018, 30, 72–83. [CrossRef]
119. Gobbi, B.; Van Rompaey, A.; Gasparri, N.I.; Vanacker, V. Forest degradation in the Dry Chaco: A detection based on 3D canopy
reconstruction from UAV-SfM techniques. For. Ecol. Manag. 2022, 526, 120554. [CrossRef]
120. Coletta, L.F.; de Almeida, D.C.; Souza, J.R.; Manzione, R.L. Novelty detection in UAV images to identify emerging threats in
eucalyptus crops. Comput. Electron. Agric. 2022, 196, 106901. [CrossRef]
121. Xiao, D.; Pan, Y.; Feng, J.; Yin, J.; Liu, Y.; He, L. Remote sensing detection algorithm for apple fire blight based on UAV
multispectral image. Comput. Electron. Agric. 2022, 199, 107137. [CrossRef]
122. Singh, P.; Pandey, P.C.; Petropoulos, G.P.; Pavlides, A.; Srivastava, P.K.; Koutsias, N.; Deng, K.A.K.; Bao, Y. Hyperspectral
remote sensing in precision agriculture: Present status, challenges, and future trends. In Hyperspectral Remote Sensing; Elsevier:
Amsterdam, The Netherlands, 2020; pp. 121–146.
123. Fuglie, K. The growing role of the private sector in agricultural research and development world-wide. Glob. Food Secur. 2016,
10, 29–38. [CrossRef]
124. Chang, A.; Yeom, J.; Jung, J.; Landivar, J. Comparison of canopy shape and vegetation indices of citrus trees derived from UAV
multispectral images for characterization of citrus greening disease. Remote Sens. 2020, 12, 4122. [CrossRef]
Drones 2023, 7, 398 39 of 42

125. Barnes, E.; Clarke, T.; Richards, S.; Colaizzi, P.; Haberland, J.; Kostrzewski, M.; Waller, P.; Choi, C.; Riley, E.; Thompson, T.;
et al. Coincident detection of crop water stress, nitrogen status and canopy density using ground based multispectral data. In
Proceedings of the Fifth International Conference on Precision Agriculture, Bloomington, MN, USA, 16–19 July 2000; Volume
1619, p. 6.
126. Gitelson, A.A.; Viña, A.; Arkebauer, T.J.; Rundquist, D.C.; Keydan, G.; Leavitt, B. Remote estimation of leaf area index and green
leaf biomass in maize canopies. Geophys. Res. Lett. 2003, 30. Available online: https://www.researchgate.net/publication/4325
6762_Coincident_detection_of_crop_water_stress_nitrogen_status_and_canopy_density_using_ground_based_multispectral_
data (accessed on 23 May 2023). [CrossRef]
127. Deng, X.; Zhu, Z.; Yang, J.; Zheng, Z.; Huang, Z.; Yin, X.; Wei, S.; Lan, Y. Detection of citrus huanglongbing based on multi-input
neural network model of UAV hyperspectral remote sensing. Remote Sens. 2020, 12, 2678. [CrossRef]
128. Kerkech, M.; Hafiane, A.; Canals, R. Deep leaning approach with colorimetric spaces and vegetation indices for vine diseases
detection in UAV images. Comput. Electron. Agric. 2018, 155, 237–243. [CrossRef]
129. Meyer, G.E.; Hindman, T.W.; Laksmi, K. Machine vision detection parameters for plant species identification. In Proceedings of
the Precision Agriculture and Biological Quality, Boston, MA, USA, 3–4 November 1999; Volume 3543, pp. 327–335.
130. Woebbecke, D.M.; Meyer, G.E.; Von Bargen, K.; Mortensen, D.A. Color indices for weed identification under various soil, residue,
and lighting conditions. Trans. ASAE 1995, 38, 259–269. [CrossRef]
131. Meyer, G.E.; Neto, J.C. Verification of color vegetation indices for automated crop imaging applications. Comput. Electron. Agric.
2008, 63, 282–293. [CrossRef]
132. Su, J.; Liu, C.; Coombes, M.; Hu, X.; Wang, C.; Xu, X.; Li, Q.; Guo, L.; Chen, W.H. Wheat yellow rust monitoring by learning from
multispectral UAV aerial imagery. Comput. Electron. Agric. 2018, 155, 157–166. [CrossRef]
133. Zhang, X.; Han, L.; Dong, Y.; Shi, Y.; Huang, W.; Han, L.; González-Moreno, P.; Ma, H.; Ye, H.; Sobeih, T. A deep learning-based
approach for automated yellow rust disease detection from high-resolution hyperspectral UAV images. Remote Sens. 2019,
11, 1554. [CrossRef]
134. Zhang, T.; Xu, Z.; Su, J.; Yang, Z.; Liu, C.; Chen, W.H.; Li, J. Ir-UNet: Irregular Segmentation U-Shape Network for Wheat Yellow
Rust Detection by UAV Multispectral Imagery. Remote Sens. 2021, 13, 3892. [CrossRef]
135. Huang, H.; Deng, J.; Lan, Y.; Yang, A.; Zhang, L.; Wen, S.; Zhang, H.; Zhang, Y.; Deng, Y. Detection of helminthosporium leaf
blotch disease based on UAV imagery. Appl. Sci. 2019, 9, 558. [CrossRef]
136. Kharim, M.N.A.; Wayayok, A.; Abdullah, A.F.; Shariff, A.R.M.; Husin, E.M.; Mahadi, M.R. Predictive zoning of pest and disease
infestations in rice field based on UAV aerial imagery. Egypt. J. Remote Sens. Space Sci. 2022, 25, 831–840.
137. Stewart, E.L.; Wiesner-Hanks, T.; Kaczmar, N.; DeChant, C.; Wu, H.; Lipson, H.; Nelson, R.J.; Gore, M.A. Quantitative phenotyping
of Northern Leaf Blight in UAV images using deep learning. Remote Sens. 2019, 11, 2209. [CrossRef]
138. Ishengoma, F.S.; Rai, I.A.; Said, R.N. Identification of maize leaves infected by fall armyworms using UAV-based imagery and
convolutional neural networks. Comput. Electron. Agric. 2021, 184, 106124. [CrossRef]
139. Simonyan, K.; Zisserman, A. Very deep convolutional networks for large-scale image recognition. arXiv 2014, arXiv:1409.1556.
140. Szegedy, C.; Vanhoucke, V.; Ioffe, S.; Shlens, J.; Wojna, Z. Rethinking the inception architecture for computer vision. In Proceedings
of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA, 27–30 June 2016; pp. 2818–2826.
141. Sandler, M.; Howard, A.; Zhu, M.; Zhmoginov, A.; Chen, L.C. Mobilenetv2: Inverted residuals and linear bottlenecks. In
Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA, 18–22 June 2018;
pp. 4510–4520.
142. Ye, H.; Huang, W.; Huang, S.; Cui, B.; Dong, Y.; Guo, A.; Ren, Y.; Jin, Y. Recognition of banana fusarium wilt based on UAV
remote sensing. Remote Sens. 2020, 12, 938. [CrossRef]
143. Tetila, E.C.; Machado, B.B.; Menezes, G.K.; Oliveira, A.d.S.; Alvarez, M.; Amorim, W.P.; Belete, N.A.D.S.; Da Silva, G.G.; Pistori, H.
Automatic recognition of soybean leaf diseases using UAV images and deep convolutional neural networks. IEEE Geosci. Remote
Sens. Lett. 2019, 17, 903–907. [CrossRef]
144. Ha, J.G.; Moon, H.; Kwak, J.T.; Hassan, S.I.; Dang, M.; Lee, O.N.; Park, H.Y. Deep convolutional neural network for classifying
Fusarium wilt of radish from unmanned aerial vehicles. J. Appl. Remote Sens. 2017, 11, 042621. [CrossRef]
145. Lu, F.; Sun, Y.; Hou, F. Using UAV visible images to estimate the soil moisture of steppe. Water 2020, 12, 2334. [CrossRef]
146. Ge, X.; Ding, J.; Jin, X.; Wang, J.; Chen, X.; Li, X.; Liu, J.; Xie, B. Estimating agricultural soil moisture content through UAV-based
hyperspectral images in the arid region. Remote Sens. 2021, 13, 1562. [CrossRef]
147. Bertalan, L.; Holb, I.; Pataki, A.; Szabó, G.; Szalóki, A.K.; Szabó, S. UAV-based multispectral and thermal cameras to predict soil
water content–A machine learning approach. Comput. Electron. Agric. 2022, 200, 107262. [CrossRef]
148. Awad, M.; Khanna, R.; Awad, M.; Khanna, R. Support vector regression. In Efficient Learning Machines: Theories, Concepts, and
Applications for Engineers and System Designers; 2015; pp. 67–80. Available online: https://www.researchgate.net/publication/27
7299933_Efficient_Learning_Machines_Theories_Concepts_and_Applications_for_Engineers_and_System_Designers (accessed
on 23 May 2023).
149. Zhang, Y.; Han, W.; Zhang, H.; Niu, X.; Shao, G. Evaluating soil moisture content under maize coverage using UAV multimodal
data by machine learning algorithms. J. Hydrol. 2023, 129086. [CrossRef]
Drones 2023, 7, 398 40 of 42

150. Zhang, X.; Yuan, Y.; Zhu, Z.; Ma, Q.; Yu, H.; Li, M.; Ma, J.; Yi, S.; He, X.; Sun, Y. Predicting the Distribution of Oxytropis
ochrocephala Bunge in the Source Region of the Yellow River (China) Based on UAV Sampling Data and Species Distribution
Model. Remote Sens. 2021, 13, 5129. [CrossRef]
151. Lan, Y.; Huang, K.; Yang, C.; Lei, L.; Ye, J.; Zhang, J.; Zeng, W.; Zhang, Y.; Deng, J. Real-Time Identification of Rice Weeds by UAV
Low-Altitude Remote Sensing Based on Improved Semantic Segmentation Model. Remote Sens. 2021, 13, 4370. [CrossRef]
152. Lu, W.; Okayama, T.; Komatsuzaki, M. Rice Height Monitoring between Different Estimation Models Using UAV Photogrammetry
and Multispectral Technology. Remote Sens. 2021, 14, 78. [CrossRef]
153. Wei, L.; Luo, Y.; Xu, L.; Zhang, Q.; Cai, Q.; Shen, M. Deep Convolutional Neural Network for Rice Density Prescription Map at
Ripening Stage Using Unmanned Aerial Vehicle-Based Remotely Sensed Images. Remote Sens. 2021, 14, 46. [CrossRef]
154. Cao, X.; Liu, Y.; Yu, R.; Han, D.; Su, B. A Comparison of UAV RGB and Multispectral Imaging in Phenotyping for Stay Green of
Wheat Population. Remote Sens. 2021, 13, 5173. [CrossRef]
155. Zhao, J.; Zhang, X.; Yan, J.; Qiu, X.; Yao, X.; Tian, Y.; Zhu, Y.; Cao, W. A wheat spike detection method in UAV images based on
improved YOLOv5. Remote Sens. 2021, 13, 3095. [CrossRef]
156. Jocher, G.; Stoken, A.; Borovec, J.; Christopher, S.; Laughing, L.C. ultralytics/yolov5: V4. 0-nn. SiLU () activations, Weights &
Biases logging, PyTorch Hub integration. Zenodo 2021. [CrossRef]
157. Wang, J.; Zhou, Q.; Shang, J.; Liu, C.; Zhuang, T.; Ding, J.; Xian, Y.; Zhao, L.; Wang, W.; Zhou, G.; et al. UAV-and Machine
Learning-Based Retrieval of Wheat SPAD Values at the Overwintering Stage for Variety Screening. Remote Sens. 2021, 13, 5166.
[CrossRef]
158. Nazeri, B.; Crawford, M. Detection of Outliers in LiDAR Data Acquired by Multiple Platforms over Sorghum and Maize. Remote
Sens. 2021, 13, 4445. [CrossRef]
159. Chen, P.; Ma, X.; Wang, F.; Li, J. A New Method for Crop Row Detection Using Unmanned Aerial Vehicle Images. Remote Sens.
2021, 13, 3526. [CrossRef]
160. Wang, F.; Yao, X.; Xie, L.; Zheng, J.; Xu, T. Rice Yield Estimation Based on Vegetation Index and Florescence Spectral Information
from UAV Hyperspectral Remote Sensing. Remote Sens. 2021, 13, 3390. [CrossRef]
161. Traore, A.; Ata-Ul-Karim, S.T.; Duan, A.; Soothar, M.K.; Traore, S.; Zhao, B. Predicting Equivalent Water Thickness in Wheat
Using UAV Mounted Multispectral Sensor through Deep Learning Techniques. Remote Sens. 2021, 13, 4476. [CrossRef]
162. Ndlovu, H.S.; Odindi, J.; Sibanda, M.; Mutanga, O.; Clulow, A.; Chimonyo, V.G.; Mabhaudhi, T. A comparative estimation of
maize leaf water content using machine learning techniques and unmanned aerial vehicle (UAV)-based proximal and remotely
sensed data. Remote Sens. 2021, 13, 4091. [CrossRef]
163. Pádua, L.; Matese, A.; Di Gennaro, S.F.; Morais, R.; Peres, E.; Sousa, J.J. Vineyard classification using OBIA on UAV-based RGB
and multispectral data: A case study in different wine regions. Comput. Electron. Agric. 2022, 196, 106905. [CrossRef]
164. Zhang, H.; Yang, W.; Yu, H.; Zhang, H.; Xia, G.S. Detecting power lines in UAV images with convolutional features and structured
constraints. Remote Sens. 2019, 11, 1342. [CrossRef]
165. Pastucha, E.; Puniach, E.; Ścisłowicz, A.; Ćwiakała,
˛ P.; Niewiem, W.; Wiacek,
˛ P. 3d reconstruction of power lines using UAV
images to monitor corridor clearance. Remote Sens. 2020, 12, 3698. [CrossRef]
166. Tan, J.; Zhao, H.; Yang, R.; Liu, H.; Li, S.; Liu, J. An entropy-weighting method for efficient power-line feature evaluation and
extraction from lidar point clouds. Remote Sens. 2021, 13, 3446. [CrossRef]
167. Zhang, Y.; Yuan, X.; Li, W.; Chen, S. Automatic power line inspection using UAV images. Remote Sens. 2017, 9, 824. [CrossRef]
168. Zhou, Y.; Xu, C.; Dai, Y.; Feng, X.; Ma, Y.; Li, Q. Dual-view stereovision-guided automatic inspection system for overhead
transmission line corridor. Remote Sensing 2022, 14, 4095. [CrossRef]
169. Ortega, S.; Trujillo, A.; Santana, J.M.; Suárez, J.P.; Santana, J. Characterization and modeling of power line corridor elements from
LiDAR point clouds. ISPRS J. Photogramm. Remote Sens. 2019, 152, 24–33. [CrossRef]
170. Zhao, Z.; Zhen, Z.; Zhang, L.; Qi, Y.; Kong, Y.; Zhang, K. Insulator detection method in inspection image based on improved
faster R-CNN. Energies 2019, 12, 1204. [CrossRef]
171. Girshick, R.; Donahue, J.; Darrell, T.; Malik, J. Rich feature hierarchies for accurate object detection and semantic segmentation.
In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Columbus, OH, USA, 23–28 June 2014;
pp. 580–587.
172. Ma, Y.; Li, Q.; Chu, L.; Zhou, Y.; Xu, C. Real-time detection and spatial localization of insulators for UAV inspection based on
binocular stereo vision. Remote Sens. 2021, 13, 230. [CrossRef]
173. Liu, C.; Wu, Y.; Liu, J.; Han, J. MTI-YOLO: A light-weight and real-time deep neural network for insulator detection in complex
aerial images. Energies 2021, 14, 1426. [CrossRef]
174. Prates, R.M.; Cruz, R.; Marotta, A.P.; Ramos, R.P.; Simas Filho, E.F.; Cardoso, J.S. Insulator visual non-conformity detection in
overhead power distribution lines using deep learning. Comput. Electr. Eng. 2019, 78, 343–355. [CrossRef]
175. Wang, S.; Liu, Y.; Qing, Y.; Wang, C.; Lan, T.; Yao, R. Detection of insulator defects with improved ResNeSt and region proposal
network. IEEE Access 2020, 8, 184841–184850. [CrossRef]
176. Wen, Q.; Luo, Z.; Chen, R.; Yang, Y.; Li, G. Deep learning approaches on defect detection in high resolution aerial images of
insulators. Sensors 2021, 21, 1033. [CrossRef]
177. Chen, W.; Li, Y.; Zhao, Z. InsulatorGAN: A Transmission Line Insulator Detection Model Using Multi-Granularity Conditional
Generative Adversarial Nets for UAV Inspection. Remote Sens. 2021, 13, 3971. [CrossRef]
Drones 2023, 7, 398 41 of 42

178. Mirza, M.; Osindero, S. Conditional generative adversarial nets. arXiv 2014, arXiv:1411.1784.
179. Liu, Z.; Miao, X.; Xie, Z.; Jiang, H.; Chen, J. Power Tower Inspection Simultaneous Localization and Mapping: A Monocular
Semantic Positioning Approach for UAV Transmission Tower Inspection. Sensors 2022, 22, 7360. [CrossRef]
180. Bao, W.; Ren, Y.; Wang, N.; Hu, G.; Yang, X. Detection of Abnormal Vibration Dampers on Transmission Lines in UAV Remote
Sensing Images with PMA-YOLO. Remote Sens. 2021, 13, 4134. [CrossRef]
181. Bao, W.; Du, X.; Wang, N.; Yuan, M.; Yang, X. A Defect Detection Method Based on BC-YOLO for Transmission Line Components
in UAV Remote Sensing Images. Remote Sens. 2022, 14, 5176. [CrossRef]
182. Nex, F.; Duarte, D.; Steenbeek, A.; Kerle, N. Towards real-time building damage mapping with low-cost UAV solutions. Remote
Sens. 2019, 11, 287. [CrossRef]
183. Yeom, J.; Han, Y.; Chang, A.; Jung, J. Hurricane building damage assessment using post-disaster UAV data. In Proceedings of the
IGARSS 2019—2019 IEEE International Geoscience and Remote Sensing Symposium, Yokohama, Japan, 28 July–2 August 2019;
pp. 9867–9870.
184. Wenzhuo, L.; Kaimin, S.; Chuan, X. Automatic 3D Building Change Detection Using UAV Images. In Proceedings of the
IGARSS 2019—2019 IEEE International Geoscience and Remote Sensing Symposium, Yokohama, Japan, 28 July–2 August 2019;
pp. 1574–1577.
185. Wu, H.; Nie, G.; Fan, X. Classification of Building Structure Types Using UAV Optical Images. In Proceedings of the IGARSS
2020—2020 IEEE International Geoscience and Remote Sensing Symposium, Waikoloa, HI, USA, 26 September–2 October 2020;
pp. 1193–1196.
186. Zheng, L.; Ai, P.; Wu, Y. Building recognition of UAV remote sensing images by deep learning. In Proceedings of the IGARSS
2020—2020 IEEE International Geoscience and Remote Sensing Symposium, Waikoloa, HI, USA, 26 September–2 October 2020;
pp. 1185–1188.
187. Li, X.; Yang, J.; Li, Z.; Yang, F.; Chen, Y.; Ren, J.; Duan, Y. Building Damage Detection for Extreme Earthquake Disaster Area
Location from Post-Event Uav Images Using Improved SSD. In Proceedings of the IGARSS 2022—2022 IEEE International
Geoscience and Remote Sensing Symposium, Kuala Lumpur, Malaysia, 17–22 July 2022; pp. 2674–2677.
188. Liu, W.; Anguelov, D.; Erhan, D.; Szegedy, C.; Reed, S.; Fu, C.Y.; Berg, A.C. SSD: Single shot multibox detector. In Proceedings of
the Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, 11–14 October 2016; Proceedings,
Part I 14; Springer: Cham, Switzerland, 2016; pp. 21–37.
189. Woo, S.; Park, J.; Lee, J.Y.; Kweon, I.S. Cbam: Convolutional block attention module. In Proceedings of the European Conference
on Computer Vision (ECCV), Munich, Germany, 8–14 September 2018; pp. 3–19.
190. Shi, X.; Huang, H.; Pu, C.; Yang, Y.; Xue, J. CSA-UNet: Channel-Spatial Attention-Based Encoder–Decoder Network for Rural
Blue-Roofed Building Extraction from UAV Imagery. IEEE Geosci. Remote Sens. Lett. 2022, 19, 1–5. [CrossRef]
191. He, H.; Yu, J.; Cheng, P.; Wang, Y.; Zhu, Y.; Lin, T.; Dai, G. Automatic, Multiview, Coplanar Extraction for CityGML Building
Model Texture Mapping. Remote Sens. 2021, 14, 50. [CrossRef]
192. Laugier, E.J.; Casana, J. Integrating Satellite, UAV, and Ground-Based Remote Sensing in Archaeology: An Exploration of
Pre-Modern Land Use in Northeastern Iraq. Remote Sens. 2021, 13, 5119. [CrossRef]
193. Ammour, N.; Alhichri, H.; Bazi, Y.; Benjdira, B.; Alajlan, N.; Zuair, M. Deep learning approach for car detection in UAV imagery.
Remote Sens. 2017, 9, 312. [CrossRef]
194. Li, J.; Chen, S.; Zhang, F.; Li, E.; Yang, T.; Lu, Z. An adaptive framework for multi-vehicle ground speed estimation in airborne
videos. Remote Sens. 2019, 11, 1241. [CrossRef]
195. Zhang, Y.; Guo, L.; Wang, Z.; Yu, Y.; Liu, X.; Xu, F. Intelligent ship detection in remote sensing images based on multi-layer
convolutional feature fusion. Remote Sens. 2020, 12, 3316. [CrossRef]
196. Lubczonek, J.; Wlodarczyk-Sielicka, M.; Lacka, M.; Zaniewicz, G. Methodology for Developing a Combined Bathymetric and
Topographic Surface Model Using Interpolation and Geodata Reduction Techniques. Remote Sens. 2021, 13, 4427. [CrossRef]
197. Ioli, F.; Bianchi, A.; Cina, A.; De Michele, C.; Maschio, P.; Passoni, D.; Pinto, L. Mid-Term Monitoring of Glacier’s Variations with
UAVs: The Example of the Belvedere Glacier. Remote Sens. 2021, 14, 28. [CrossRef]
198. Nardin, W.; Taddia, Y.; Quitadamo, M.; Vona, I.; Corbau, C.; Franchi, G.; Staver, L.W.; Pellegrinelli, A. Seasonality and
Characterization Mapping of Restored Tidal Marsh by NDVI Imageries Coupling UAVs and Multispectral Camera. Remote Sens.
2021, 13, 4207. [CrossRef]
199. Kim, M.; Chung, O.S.; Lee, J.K. A Manual for Monitoring Wild Boars (Sus scrofa) Using Thermal Infrared Cameras Mounted on
an Unmanned Aerial Vehicle (UAV). Remote Sens. 2021, 13, 4141. [CrossRef]
200. Rančić, K.; Blagojević, B.; Bezdan, A.; Ivošević, B.; Tubić, B.; Vranešević, M.; Pejak, B.; Crnojević, V.; Marko, O. Animal Detection
and Counting from UAV Images Using Convolutional Neural Networks. Drones 2023, 7, 179. [CrossRef]
201. Ge, S.; Gu, H.; Su, W.; Praks, J.; Antropov, O. Improved semisupervised unet deep learning model for forest height mapping with
satellite sar and optical data. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2022, 15, 5776–5787. [CrossRef]
202. Zhang, B.; Ye, H.; Lu, W.; Huang, W.; Wu, B.; Hao, Z.; Sun, H. A spatiotemporal change detection method for monitoring pine
wilt disease in a complex landscape using high-resolution remote sensing imagery. Remote Sens. 2021, 13, 2083. [CrossRef]
203. Barrile, V.; Simonetti, S.; Citroni, R.; Fotia, A.; Bilotta, G. Experimenting Agriculture 4.0 with Sensors: A Data Fusion Approach
between Remote Sensing, UAVs and Self-Driving Tractors. Sensors 2022, 22, 7910. [CrossRef] [PubMed]
Drones 2023, 7, 398 42 of 42

204. Zheng, Q.; Huang, W.; Cui, X.; Shi, Y.; Liu, L. New spectral index for detecting wheat yellow rust using Sentinel-2 multispectral
imagery. Sensors 2018, 18, 868. [CrossRef] [PubMed]
205. Bohnenkamp, D.; Behmann, J.; Mahlein, A.K. In-field detection of yellow rust in wheat on the ground canopy and UAV scale.
Remote Sens. 2019, 11, 2495. [CrossRef]
206. Saeed, Z.; Yousaf, M.H.; Ahmed, R.; Velastin, S.A.; Viriri, S. On-Board Small-Scale Object Detection for Unmanned Aerial Vehicles
(UAVs). Drones 2023, 7, 310. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.