An Improved Yolov5 Marine Biological Object Detection Algorithm

YOLO algorithm has high real-time monitoring speed and average accuracy, and also has great advantages for target detection in complex Marine environment. The research of the algorithm must be applied to the equipment eventually, but in most cases, the storage capacity of the equipment is limited, and at the same time, it needs to meet the requirements of high-precision detection. Therefore, this paper proposes an improved Marine biometrics algorithm for YOLOv5 network, which uses GhostNet's idea and introduces GhostBottleneck into YOLOv5. It can be used as a plug and play module to upgrade the existing convolutional neural network. It can reduce the computation of the network and ensure the precision of the network. On this basis, CBAM module is introduced, which combines spatial attention mechanism and channel attention mechanism, and uses multiscale maximum pooling layer to increase the range of receptive field, which can significantly improve the accuracy of image classification and target detection. The experimental results show that compared with the original YOLOv5, the improved model occupies much less storage space and has a greater improvement in the identification accuracy of Marine organisms.