Fast detection of student classroom behavior based on CBW-YOLOv5s lightweight model
To address the issues of low accuracy and high model complexity in existing student behavior detection algorithms,a lightweight student behavior recognition algorithm based on CBW-YOLOv5s(CA-BiFPN-WIoU)is proposed.The whole algo-rithm adopts the lightweight structure of GhostNet to ensure detection performance while simplifying the network structure,thus sig-nificantly reducing the model complexity.The PANet is replaced by BiFPN to enhance the model's ability to extract and fuse features of student behavior.CA attention mechanism is integrated into some C3 modules to enable the network to pay attention to the spatial and channel information of feature layers.The Wise-IoU loss function is used to improve the model's performance.The experimental results show that the proposed method effectively simplifies the model complexity and maintains a high detection accuracy.
万方数据
