AI-enabled Object Detection in UAVs: Challenges, Design Choices, and Research Directions
Ayush Jain*1
Rohit Ramaprasad*1
Dr. Pratik Narang1
Dr. Murari Mandal4
Dr. Vinay Chamola1
F. Richard Yu3
Mohsen Guizani4
1Bits Pilani 2National University of Singapore 3Carleton University 4Qatar University
IEEE Network
The authors are grateful to IBM for providing online access to the Power9 AC922 GPU server.
Introduction
Unmanned Aerial Vehicles (UAVs) are emerging as a powerful tool for various industrial and smart city applications. The UAVs coupled with various sensors can perform many cognitive tasks such as object detection, surveillance, traffic management, and urban planning. Deep learning has emerged as a popular technique to speed up the processing of high dimensional data like images and videos which has led to several applications in surveillance and autonomous driving. However, the area of aerial object detection has been understudied. This work proposes a deep learning approach for detection of objects in aerial scenes captured by UAVs. Our work first categorizes the current methods for aerial object detection using deep learning techniques and discusses how the task is different from general object detection scenarios. We delineate the specific challenges involved and experimentally demonstrate the key design decisions which significantly affect the accuracy and robustness of models. We further propose an optimized architecture which utilizes these optimal design choices along with the recent ResNeSt backbone to achieve superior performance in aerial object detection. Lastly, we propose several research directions to inspire further advancement in aerial object detection.
Publication
IEEE Network.Paper | Code
If you find our project useful, please consider citing us:
@article{jain2021a,
author = {Jain, A. and Ramaprasad, R. and Narang, P. and Mandal, M. and Chamola, V. and Yu, F.Richard and Guizani, M.},
title = {AI-enabled Object Detection in UAVs: Challenges, Design Choices, and Research Directions"},
date = {2021-03},
language = {en},
journal = {IEEE Network}
}