Research on Intelligent Management of Fishing Ground Based on Target Detection

Object detection technique is adopted in fishery, that is detecting fish. We choose to use YOLO v4, a new and efficient target detection network for sampling detection. Manual counting of fish is prone to deviations and costs a lot of manpower. Using YOLO v4 to detect fish can improve the working efficiency of the fishery and reduce management costs. We used the pictures of fishes taken from the videos taken in the pipeline of Qiandao Lake fishery. Then we preprocessed the pictures, enriched the data set and use them for training to achieve the engineering of fish detection. Finally, the target recognition accuracy of the training is over 85%, and the fps is over 14. It can realize the function of real-time detection on the basis of accurately and accurately detecting fish.