Object Detection for Facial and Tongue Regions Based on Improved YOLOv10

With the rapid development of intelligent traditional Chinese medicine (TCM), the application of object detection technology to achieve face and tongue detection holds significant importance for advancing the digitalization and intelligence of TCM. To address the challenges of high detection difficulty and low accuracy in existing object detection technologies for face and tongue detection, an improved YOLOv10-based object detection algorithm named YOLOv10-GC is proposed. This algorithm enhances the C2f module by integrating the GhostNetV2 network, introducing the C2f_GhostV2 module, and incorporates the CBAM attention mechanism, achieving lightweight model design while improving feature extraction capabilities. Experimental results demonstrate that compared to the baseline model YOLOv10n, YOLOv10-GC improves [email protected] by 4.1%, reaching 85.3%, while reducing model weight, parameter count, and computational load. Additionally, it increases FPS from 164 to 200. Furthermore, compared to mainstream YOLO models, YOLOv10-GC exhibits higher detection accuracy and efficiency.