Underwater biological target detection method based on YOLOv8 model

The invention discloses an underwater biological target detection method based on a YOLOv8 model, and belongs to the field of underwater biological detection, and the method comprises the steps: fusing a channel attention mechanism and a space attention mechanism to obtain a residual attention mechanism; the bidirectional feature pyramid network and the residual attention mechanism are embedded into the YOLOv8 model, and an optimized YOLOv8 model is obtained; the optimized YOLOv8 model is trained and verified through the preprocessed underwater biological image, and a target detection model is obtained; and detecting the real-time shot image through the target detection model to obtain an underwater biological detection result. According to the method, the feature extraction capability of the backbone network is enhanced by embedding the residual attention mechanism into the YOLOv8 model, the problems of false detection and missing detection can be effectively solved, the bidirectional feature pyramid network