Abstract: In order to solve the problems of high labor cost and heavy workload in detection and identification of fish in the electronic monitoring data of traditional commercial fishing vessels, an improved detection and identification method of fish in the electronic monitoring data of Yolov8 commercial fishing vessels was employed. In this method, the GCBlock structure was used in the backbone network to model the remote dependency relationship and increase the ability of feature extraction; a novel GSConv convolution method was used at the Neck terminal to reduce the network computation; SIOU loss function was used to solve the limitation of CIOU loss function and improve the network detection accuracy. The results showed that the mAP@0.5 of the proposed Yolov8n-GCBlock-GSConv method on different labels (L1 and L2) of FishNet dataset was 43.6% and 52.7%, respectively, which was 2.0% and 4.3% higher than that of the original Yolov8n model. The computational complexity is 7.7 GFLOPS, which is 0.5 GFLOPS lower than that of the original model. The research data indicate that the Yolov8n-GCBlock-GSConv network model can quickly and accurately detect and identify fish in the electronic monitoring data of commercial fishing vessels with lower cost.
袁红春, 陶磊. 基于改进的Yolov8商业渔船电子监控数据中鱼类的检测与识别[J]. 大连海洋大学学报, 2023, 38(3): 533-542.
YUAN Hongchun, TAO Lei. Detection and identification of fish in electronic monitoring data of commercial fishing vessels based on improved Yolov8. Journal of Dalian Ocean University, 2023, 38(3): 533-542.
2023, Vol. 38 
