Methods studies for attached marine organisms detecting based on convolutional neural network

Abstract

The use of neural networks to detect marine targets has been an important research direction in the field of target detection. Traditional detection models are generally only for fish, coral, large plants and animals and other easy to detect targets, for the attached marine species that are not easy to detect there are problems such as lower detection accuracy, slower speed, and large interference in the sea environment. With the increasing problems caused by the attachment of marine organisms, the identification methods for marine attached organisms are also receiving more and more attention. In this paper, the original YOLO v5 is improved in terms of activation function, optimized image information extraction method, and balanced positive and negative samples. Finally, an improved YOLO v5 target detection algorithm is proposed and tested on a self-built dataset. The experimental results show that the values of mAP and F1 of the improved YOLO v5 target detection algorithm are 72.1% and 0.722, respectively, which are better than other target detection algorithms in terms of accuracy and reliability.