Convolution and Swim Transformer-Based Object Detection in Remote Sensing

Remote sensing object recognition in natural settings presents significant challenges, particularly with small targets and intricate backdrops. The RAST-YOLO algorithm addresses these challenges by combining the Region Attention (RA) mechanism with the Swin Transformer backbone, enhancing feature extraction for better detection accuracy. Moreover, the C3D module effectively merges deep and shallow semantic information, tackling the multi-scale problem in remote sensing targets and improving detection accuracy, especially for tiny objects. Extensive testing on the DIOR and TGRS-HRRSD datasets demonstrates the algorithm's superior robustness, accuracy, and efficiency. When compared to baseline networks, RAST-YOLO shows a notable improvement in mean average accuracy (mAP) on both datasets. Additionally, research using methods like YOLOv5x6 and YOLOv8 reveals promising results; YOLOv5x6, for instance, achieves a significant mAP enhancement of over 0.80%, highlighting its suitability for advanced remote sensing object detection applications.