CRAS-YOLO: A Novel Multi-Category Vessel Detection and Classification Model Based on YOLOv5s Algorithm

Multi-category vessel detection and classification based on satellite imagery attract a lot of attention due to their significant applications in the military civilian domains. In this study, we generated new Artificial-SAR-Vessel dataset combination FUSAR-Ship SimpleCopyPaste method. We further proposed novel multi-category called CRAS-YOLO which consisted convolutional block module (CBAM), receptive fields (RFB), adaptively spatial feature fusion (ASFF) YOLOv5s. The improved pyramid network path aggregation (PANet), integrates RFB enhancement ASFF strategy obtain richer information realize adaptive multi-scale features (RA-PANet). At same time, CBAM is added backbone accurately locate location improve capability. results confirmed that model reached precision, recall rate, mean average precision (mAP) (0.5) up 90.4%, 88.6%, 92.1% respectively. also outperformed previous studies’ another Sar Ship Detection (SSDD) with recall, mAP scores 97.3%, 95.5%, 98.7%