Aiming at the problem of insufficient storage space and limited computing ability of underwater mobile devices, an underwater garbage detection algorithm based on an improved YOLOv5s algorithm is proposed. The algorithm replaces the feature extraction module of the YOLOv5s network with the lightweight network MobileNetv3; the Convolutional Block Attention Module (CBAM) is embedded in the network to improve the feature extraction ability of the network in two dimensions of space and channel. At the same time, the improved network is pruned to reduce the redundant parameters and further compress the model. The experimental results show that the detection accuracy of the approach can reach 97.5% based on one-ninth of the parameters of YOLOv5s, and the real-time detection speed on the CPU is 2.5 times that of YOLOv5s.

针对水下移动设备存储空间不足、计算能力有限的问题，提出一种基于改进的YOLOv5s算法的水下垃圾检测算法。该算法将YOLOv5s网络的特征提取模块替换为轻量级网络MobileNetv3；将卷积块注意模块（CBAM）嵌入网络中，以提高网络在空间和通道两个维度上的特征提取能力。同时对改进后的网络进行剪枝，减少冗余参数，进一步压缩模型。实验结果表明，基于YOLOv5s的九分之一参数，该方法的检测准确率可以达到97.5%，在CPU上的实时检测速度是YOLOv5s的2.5倍。