IEEE Geoscience and Remote Sensing Magazine - March 2022 - 97

from the original high-resolution images and learning
them through the use of a CNN network without degrading
the resolution.
Alternately, the flexible movement of the camera results
in an uneven density of captured objects. Tightly packed
objects in an image, especially smaller-size ground objects,
inevitably overlap. Mekhalfi et al. [88] introduced capsule
networks to model the relationships among objects.
OBJECT DETECTION WITH DIRECTION DIVERSITY
Objection direction from an optical RS image is related to
its actual parking location. The classic CNNs, which benefit
from using a rectangular convolution kernel, are sensitive to
object direction. Figure 5(b) shows four commonly used solutions
based on DL.
The simplest and most common solution is data augmentation,
which can make CNNs rotation invariant
TABLE 1. DL-BASED SOD APPROACHES FOR UAV-EXCLUSIVE METHODS.
REFERENCE
CHALLENGES
RRNet [53]
SlimYOLOv3 [65]
Zhang et al. [62]
FS-SSD [66]
SAMFR [67]
ClusDet [57]
CenterNet [58]
Yang et al. [59]
Wu et al. [68]
NDFT [69]
MSOA-Net [56]
GDF-Net [61]
HRDNet [70]
D-A-FS SSD [63]
UAV-YOLO [16]
SyNet [71]
ComNet [72]
MPFPN [73]
D2Det [74]
DAGN [75]
GANet [76]
DAN [77]
Zhang et al. [78]
DNOD [79]
ECascade-RCNN [54]
Cas_RCNN + FPN [80]
DSYolov3 [55]
DSHNet [81]
Small objects and
scale variation
Real time
Small objects
Small objects
Scale variation
Scale variation
Scale variation
Scale variation
Real time
UAV-specific nuisances
Scale
variation
Scale variation
Scale variation
Scale variation
Small scale
Class imbalance
Blurred edges
and low contrast
Small objects
Localization
and classification
Small objects
Small objects
Dense distribution
and small objects
Real time
Dense objects and
small objects
Scale variation
Cost
Scale variation
Long-tail
distribution
DATA SET USED
VisDrone
VisDrone
VisDrone
Stanford Drone
VisDrone
VisDrone and UAVDT
VisDrone
Stanford Drone
CARPK
VisDrone and UAVDT
UVSD
VisDrone and UAVDT
VisDrone
VisDrone
UAV123 and own
VisDrone
Own
VisDrone
UAVDT
VEDAI
UAVDT, CARPK, and
PUCPR+
VisDrone-DET
Stanford Drone
VisDrone-DET and
UAVDT
VisDrone
VisDrone
VisDrone and UAVDT
ICCV
ICCV Workshop
IEEE Transactions on Circuits Systems
for Video Technology
ICCV Workshop
ICCV
-
IEEE Access
DDCLS
ICCV
Remote Sensing
Remote Sensing
CVPR
ICIT
Sensors
ICPR
IEEE Transactions on Geoscience
and Remote Sensing
IEEE Access
CVPR
IEEE Geoscience and Remote
Sensing Letters
ACM-MM
NCC
Neurocomputing
Neurocomputing
ICARA
Transportation Research Record
Journal of Visual Communication
and Image Representation
VisDrone and UAVDT WACV
JOURNAL/CONFERENCE
ICCV Workshop
YEAR
2019
2019
2019
2019
2019
2019
2019
2019
2019
2019
2020
2020
2020
2020
2020
2020
2020
2020
2020
2020
2020
2020
2020
2021
2021
2021
2021
2021
CODE LINK
https://github.com/ouc
-ocean-group/RRNet
https://github.com/
PengyiZhang/SlimYOLOv3
-
-
-
https://github.com/
fyangneil
-
-
-
https://github.com/
VITA-Group/UAV-NDFT
-
-
-
-
-
https://github.com/
mertalbaba/SyNet
-
-
https://github.com/
JialeCao001/D2Det
-
https://isrc.iscas.ac.cn/
gitlab/research/ganet
-
-
-
-
-
-
https://github.com/
we1pingyu/DSHNet
ACM: Association for Computing Machinery; ACM-MM: ACM International Conference on Multimedia; CARPK: Car Parking Lot; CVPR: IEEE Conference on Computer Vision and Pattern
Recognition; DDCLS: IEEE Data Driven Control and Learning Systems Conference; DSHNET: dual sampler and head detection network; ICARA: International Conference on Automation,
Robotics, and Applications; ICCV: International Conference on Computer Vision; ICIT: International Conference on Information Technology; ICPR: International Conference on Pattern
Recognition; NCC: National Conference on Communications; NDFT: nuisance disentangled feature transform; PUCPR+: Pontifical Catholic University of Parana+ Dataset; UAVDT: UAV
Detection and Tracking data set; UVSD: UAV-Based Vehicle Segmentation data set; VEDAI: Vehicle Detection in Aerial Imagery data sets; WACV: IEEE Winter Conference on Applications
of Computer Vision.
MARCH 2022 IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE
97
IEEE Geoscience and Remote Sensing Magazine - March 2022

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - March 2022
