AirTrack: Onboard Deep Learning Framework for Long-Range Aircraft Detection and Tracking
AirTrack: Onboard Deep Learning Framework for Long-Range Aircraft Detection and Tracking
Sourish Ghosh, Jay Patrikar, Brady Moon, Milad Moghassem Hamidi, and Sebastian Scherer
Paper Link: https://arxiv.org/pdf/2209.12849.pdf
This paper introduces, AirTrack, a real-time vision-only detect and tracking framework that respects the size, weight, and power (SWaP) constraints of sUAS systems. Given the low Signal-to- Noise ratios (SNR) of far away aircraft, we propose using full resolution images in a deep learning framework that aligns successive images to remove ego-motion. The aligned images are then used downstream in cascaded primary and secondary classifiers to improve detection and tracking performance on multiple metrics. Empirical evaluations show that our system has a probability of track of more than 95% up to a range of 700m.
0:00 - Introduction
0:42 - AirTrack Framework
1:33 - Qualitative Results
1:46 - Field Experiments
2:30 - Quantitative Results
@sourishghosh6940 @bradygmoon @jaypatrikar8183