Super accurate low latency object detection on a surveillance UAV
This addresses the need for real-time object tracking on drones in security and surveillance, though it appears incremental as it builds on existing optimization techniques.
The paper tackles the problem of achieving both high accuracy and low latency for object detection on surveillance UAVs, proposing a multi-dataset learning strategy and optimization steps that result in a more than 10x speed boost on embedded hardware with negligible accuracy loss.
Drones have proven to be useful in many industry segments such as security and surveillance, where e.g. on-board real-time object tracking is a necessity for autonomous flying guards. Tracking and following suspicious objects is therefore required in real-time on limited hardware. With an object detector in the loop, low latency becomes extremely important. In this paper, we propose a solution to make object detection for UAVs both fast and super accurate. We propose a multi-dataset learning strategy yielding top eye-sky object detection accuracy. Our model generalizes well on unseen data and can cope with different flying heights, optically zoomed-in shots and different viewing angles. We apply optimization steps such that we achieve minimal latency on embedded on-board hardware by fusing layers, quantizing calculations to 16-bit floats and 8-bit integers, with negligible loss in accuracy. We validate on NVIDIA's Jetson TX2 and Jetson Xavier platforms where we achieve a speed-wise performance boost of more than 10x.