Underwater biological target recognition of East Juyanhai based on image enhancement and yolov5

This study uses aquatic species as its research subject and the East Juyanhai of the Heihe River as its research area. We collected photos of aquatic species in the East Juyanhai, built a dataset using technical means such as underwater video acquisition and image enhancement recognition technology, and pre-processed the dataset by automatic color scale algorithm and data augmentation technology to address the issue that it is difficult to conduct out target detection of aquatic organisms in the East Juyanhai. The yolov5 algorithm is modified, and the CBAM attention mechanism is implemented to improve the detection accuracy, in order to address the issue that underwater photos are blurry and the feature extraction capacity is poor owing to the huge size difference. The results of the experiments demonstrate that the suggested algorithm for picture enhancement and improvement may successfully recognize underwater biometrics with a detection accuracy of 95%. This study can further examine the contributing reasons to changes in aquatic organisms and give data and technical references for the preservation and restoration of the water ecology in the East Juyanhai by combining the recognition findings with complete environmental data.