A FPN-Based Framework for Vehicle Detection in Aerial Images

Abstract

Vehicle detection in aerial images plays an important role in many aspects, including traffic surveillance, urban planning, parking lot analysis, etc. However, due to the influence of low resolution, complex background and rotating objects, vehicle detection in aerial images still has limited progress. In this paper, we propose a specialized framework for vehicle detection in aerial images. In this framework, Feature Pyramid Network and Faster RCNN are combined to utilize both low-resolution, semantically strong features and high resolution, semantically weak features to achieve a better detection rate of small objects. On this basis, focal loss function is adopted to reduce the imbalance between easy and hard samples. The experiments show that the performance of proposed framework is better than other state-of-the-art frameworks.