基于深度学习的森林巡检目标检测与追踪研究OA

In view of the problems existing in the traditional inspection methods of artificial forests,such as limited inspection range,high risk factor and low efficiency.To effectively solve these problems,a real-time forest inspection system based on the improved YO-LOv8n algorithm is proposed.The improved GAM-YOLOv8n attention mechanism significantly enhances the feature perception and loca-tion ability of complex targets in forest environment monitoring by optimizing the 3D arrangement of channel attention and the two-layer MLP structure,as well as the double-convolutional layer design of spatial attention and the grouped convolution strategy,thereby improv-ing the recognition accuracy of the system for situations such as fire hazards.The system integrates infrared thermal imager technology to effectively solve the problem of target detection in the case of poor visual effects in forest environments,ensuring the accuracy and timeli-ness of inspection work.After testing,the improved GAM-YOLOv8n model outperforms the YOLOv8n algorithm in key indicators such as accuracy,recall rate,and average accuracy,fully verifying that this model has excellent recognition accuracy and working efficiency in re-al and complex forest inspection scenarios.