Abstract

We propose an approach to detect flflying objects such as UAVs and aircrafts when they occupy a small portion of the fifield of view, possibly moving against complex backgrounds, and are fifilmed by a camera that itself moves. Solving such a diffificult problem requires combining both appearance and motion cues. To this end we propose a regression-based approach to motion stabilization of local image patches that allows us to achieve effective classifification on spatio-temporal image cubes and outperform stateof-the-art techniques. As the problem is relatively new, we collected two challenging datasets for UAVs and Aircrafts, which can be used as benchmarks for flflying objects detection and visionguided collision avoidance.