Enhanced Two-stage Ship Detection Algorithm in Remote Sensing Image
Tengfei Qiao, Chao Zhang, Jianping Hu, Yu‐Wen Chen · 2022
Remote sensing images have the characteristics of extreme high resolution, small object and sparse distribution., which bring huge difficulties for ship detection in the sea. Traditional object detection models based on deep learning can not be directly applied to remote sensing images. This paper proposes an efficient ship detection framework based on Cascade R-CNN network. We use the YOLOv5 model as the first stage detection and simplify the Cascade R-CNN detector head. This new model is particularly suitable for detecting sparse objects in high resolution image.