Flying objects detection from a single moving camera

Abstract

We propose an approach to detect flying objects such as UAVs and aircrafts when they occupy a small portion of the field of view, possibly moving against complex backgrounds, and are filmed by a camera that itself moves.Solving such a difficult 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 classification 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 flying objects detection and visionguided collision avoidance.

Detection of aircraft below the horizon for vision‐based detect and avoid in unmanned aircraft systems

“…These collision threats are unlikely to be detected by systems that search for aircraft with large apparent motion between two consecutive frames. Recently, progress has also been made in detecting aircraft from their visual appearance (rather than motion) by training convolutional neural networks (CNNs) to recognize aircraft in unstabilized image sequences . Although this appearance‐based approach is promising, its performance is fundamentally limited by the availability of training data (which is currently extremely limited and difficult to collect for all classes of aircraft).…”

“…Recently, progress has also been made in detecting aircraft from their visual appearance (rather than motion) by training convolutional neural networks (CNNs) to recognize aircraft in unstabilized image sequences. 19 Although this appearance-based approach is promising, its performance is fundamentally limited by the availability of training data (which is currently extremely limited and difficult to collect for all classes of aircraft). Detection approaches that search for aircraft based on their visual motion characteristics therefore remain an important cornerstone of vision-based aircraft detection, although there is a strong case for examining approaches that detect small motions over multiple consecutive frames.…”

Detecting Flying Objects Using a Single Moving Camera

“…Both CNN and Boosted trees methods allow us to outperform state-of-the-art techniques on two challenging datasets. The CNN proved to be more suitable for motion compensation than the Boosted trees introduced in our previous work [15].…”

“…We first proposed using st-cubes for flying objects detection in an earlier conference paper [15]. In this initial version of our processing pipeline, we performed motion compensation using boosted trees.…”

Collision avoidance based on line-of-sight angle

“…(see [4]- [6] and reference therein). Nevertheless, the problem of detecting a possible imminent collision is beyond the scope of this paper.…”