Abstract
Commercial Unmanned aerial vehicle (UAV) industry, which is publicly known as drone, has seen a tremendous increase in last few years, making these devices highly accessible to public. This phenomenon has immediately raised security concerns due to fact that these devices can intentionally or unintentionally cause serious hazards. In order to protect critical locations, the academia and industry have proposed several solutions in recent years. Computer vision is extensively used to detect drones autonomously compared to other proposed solutions such as RADAR, acoustics and RF signal analysis thanks to its robustness. Among these computer vision-based approaches, we see the preference of deep learning algorithms thanks to their effectiveness. In this paper, we are presenting an autonomous drone detection and tracking system which uses a static wide-angle camera and a lower-angle camera mounted on a rotating turret. In order to use memory and time efficiently, we propose a combined multi-frame deep learning detection technique, where the frame coming from the zoomed camera on the turret is overlaid on the wide-angle static camera’s frame. With this approach, we are able to build an efficient pipeline where the initial detection of small sized aerial intruders on the main image plane and their detection on the zoomed image plane is performed simultaneously, minimizing the cost of resource exhaustive detection algorithm. In addition to this, we present the integral system including tracking algorithms, deep learning classification architectures and the protocols.
Highlights
The exponentially increasing public accessibility of drones has been posing a great threat to the general security and confidentiality
We present an autonomous drone detection, tracking and identification system based on optics and deep learning, composed of a static wide-angle RGB camera platform and a rotating turret, where a lower-angle RGB camera is mounted
For tracking purpose, where the intruding airborne target is being tracked on wide-angle image plane, and verified by rotating zoom cameras; we have introduced a novel algorithm called Target Candidate Track (TCT)
Summary
The exponentially increasing public accessibility of drones has been posing a great threat to the general security and confidentiality. The static wideangle camera serves as a primary aerial object detection, where drones can be detected at relatively long range (up to ∼ 1 km), even as small as few dozens of pixels These detections are tracked on the image plane of the wide camera and the ones which show specific motion and visual signatures are inspected by the narrow-angle RGB camera on the rotating turret. For detection of possible drones on the wide-angle camera’s image plane, a lightweight version of YOLO deep learning algorithm is used, which has recently become a popular choice thanks to its robustness and speed [14] This lightweight architecture is extensively trained for the detection of drones, as small as 6 × 6 pixels, for backgrounds similar to the operational one. We have decided to use a Kalman tracking algorithm based on an already existing python library, optimized for algebraic operations [36]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: IPSJ Transactions on Computer Vision and Applications
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
