A Cross Stage Partial Network with Strengthen Matching Detector for Remote Sensing Object Detection

Remote sensing object detection is a difficult task because it often requires real-time feedback through numerous objects in complex environments. In detection, Feature Pyramids Networks (FPN) have been widely used for better representations based on multi-scale problem. However, the multiple level features cause detectors’ structures to be and makes redundant calculations that slow down detector. This paper uses single-layer feature make lightweight accurate without relying Pyramid Structures. We proposed method called Cross Stage Partial Strengthen Matching Detector (StrMCsDet). The StrMCsDet generates single-level map architecture backbone with cross stage partial network. To provide an alternative way of replacing traditional pyramid, encoder was designed compensate receptive field limitation. Additionally, stronger matching strategy sure various scale anchors may equally matched. different from conventional full pyramid structure fully exploits which deals encoder. Methods achieved both comparable precision speed practical applications. Experiments conducted DIOR dataset NWPU-VHR-10 65.6 73.5 mAP 1080 Ti, respectively, can match performance state-of-the-art works. Moreover, less computation 38.5 FPS dataset.