Vehicle Detection in Very-High-Resolution Remote Sensing Images Based on an Anchor-Free Detection Model with a More Precise Foveal Area

Vehicle detection in aerial images is a challenging task. The complexity of the background information and redundancy area are main obstacles that limit successful operation vehicle based on anchors very-high-resolution (VHR) remote sensing images. In this paper, an anchor-free target method proposed to solve problems above. First, multi-attention feature pyramid network (MA-FPN) was designed address influence noise by fusing attention (FPN) structure. Second, more precise foveal (MPFA) provide better ground truth for determining accurate positive sample selection area. model with MA-FPN MPFA can predict vehicles accurately quickly VHR through direct regression pixels map. A detailed evaluation image (RSI) imagery (VEDAI) data sets shows our performs well, simple, fast.