Underwater target detection is of great significance in marine scienceenvironmental protectionresource developmentmilitary defensecultural heritage protection and other fields.Howeverthe complex underwater environmentpoor underwater image quality and small biological aggregation may lead to missed detection and false detection in underwater target detectionso it is necessary to improve the detection accuracy.The realtime detection needs to design a faster network structure.Underwater devices have limited storage and computing power and need to maintain low computational overhead while ensuring accuracy.In view of these difficultiesan improved network YOLOv7PSS is proposed based on YOLOv7.FirstlyPConv convolution is used to replace some ordinary convolutions in the backbone network to reduce parameter quantity of the model and speed up training and prediction of the model.Thenthe SE attention mechanism is added to enhance the feature extraction abilityand SIoU loss function is adopted to accelerate network convergence and optimize model training process.Experimental results show that on the URPC2021 underwater target detection datasetthe proposed algorithm has a mAP of 87.3%which is 7.5% higher than that of the original modeland the parameter quantity is reduced by 11.9%which lays a foundation for the deployment of underwater equipment.