Tackling the Background Bias in Sparse Object Detection via Cropped Windows
This work addresses a domain-specific challenge in remote sensing for UAV applications, with incremental improvements in detection capability.
The paper tackles the problem of sparse object detection in UAV recordings by proposing a tiling method to reduce background bias and enable higher image resolutions, resulting in substantial performance improvements validated on three datasets.
Object detection on Unmanned Aerial Vehicles (UAVs) is still a challenging task. The recordings are mostly sparse and contain only small objects. In this work, we propose a simple tiling method that improves the detection capability in the remote sensing case without modifying the model itself. By reducing the background bias and enabling the usage of higher image resolutions during training, our method can improve the performance of models substantially. The procedure was validated on three different data sets and outperformed similar approaches in performance and speed.