Identification And Research Of Illegal Acts In The Yangtze River Arrest Prohibition

As the longest river in China and one of the largest rivers in the world,the Yangtze River basin covers 11 provinces and cities in China,with rich fishery resources.However,with population growth and economic development,the protection and management of fishery resources in the Yangtze River has become increasingly important.Especially in the past few decades,due to factors such as overfishing,water pollution,and river management,the Yangtze River fishery resources have faced tremendous pressure,and the implementation of the fishing ban policy has directly affected the restoration and protection of the Yangtze River fishery resources.Currently,the identification of illegal activities in prohibited fishing areas mainly relies on manual patrols and video surveillance,which consumes time,manpower,and material resources,and there are issues such as missing reports and false reports.Therefore,using computer vision and machine learning technology to develop an automatic identification system for illegal activities in closed fishing areas has important practical significance for the protection and management of fishery resources in the Yangtze River.This paper mainly discusses human motion detection methods based on posture attention mechanism.Traditional human motion detection methods mostly use CNN(convolutional neural network)for feature extraction and classification,but their detection accuracy for human motion in different poses is relatively low.Therefore,this paper proposes an improved YOLOv5 behavior recognition algorithm based on attention mechanism.Specifically,this article uses the YOLOv5 algorithm as the foundation and proposes an instance centered attention mechanism.Aiming at the shortcomings of this mechanism,this paper proposes a human posture based attention mechanism combined with key point detection methods,and analyzes the shortcomings of this mechanism in COCO data sets.In order to further optimize the posture attention mechanism,this article adds channels and spatial attention mechanisms,and conducts comparative experiments with existing algorithms to verify.The results show that compared to other attention modules,the CBAM module is more suitable for tasks related to human motion detection after strengthening features.In order to improve the pertinence of the public dataset,this article adds 3000 datasets taken in the Yangtze River waters to the proposed method model.Its feature is to make a comparison diagram of the brightness and distance of the photos.The results show that the method achieves excellent performance on this dataset.At the same time,this article also introduces the key point detection,target detection,dense posture estimation and other tasks in human motion detection in detail.The experimental results in this paper verify the effectiveness and superiority of the proposed method in human posture detection.These methods can be used in many human motion detection related tasks.These technologies have broad application prospects in human behavior recognition,motion control,and other fields.Future work includes validating the methods proposed in this article on more datasets and further optimizing the model to improve its performance.