通信与信息技术Issue(2)：21-25,5.
基于深度学习的森林巡检目标检测与追踪研究
Research on forest inspection target detection and tracking based on deep learning
摘要
Abstract
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.关键词
森林巡检/深度学习/红外热像仪/YOLOv8n/GAM/ByteTrackKey words
Forest inspection/Deep learning/Infrared thermal imager/YOLOv8n/GAM/ByteTrack分类
信息技术与安全科学引用本文复制引用
黄银旭,喻恒,邵孟轩,李圣普,王艺馨..基于深度学习的森林巡检目标检测与追踪研究[J].通信与信息技术,2026,(2):21-25,5.基金项目
平顶山学院大学生创新创业训练项目(项目编号:109192025048)河南省科技攻关项目(项目编号:232102210004)河南省高等学校科学技术研究重点项目(项目编号:23A520041) （项目编号:109192025048）
