基于改进YOL0v7的电力施工现场异常行为检测分析
Analysis of Abnormal Behavior Detection in Electric Power Construction Sites Based on Improved YOLOv7
吴宛凝 1王莉1
作者信息
- 1. 南京工业大学电气工程与控制科学学院,江苏 210032
- 折叠
摘要
阐述一种基于改进YOLOv7算法的电力施工现场人员异常行为检测方法,该方法在head区域增加SE注意力机制,增强特征提取能力,同时增加小目标专用检测头,提高在大场景小目标环境下的识别能力.实验结果表明,改进后的模型精确率提高0.7％,mAP提高2.1％,达到94.5％,从而实现对电力施工现场异常行为检测,特别是小目标的实时检测.
Abstract
This paper expounds a method for detecting abnormal behavior of personnel in power construction sites based on the improved YOLOv7 algorithm.The method adds SE attention mechanism in the head area to enhance feature extraction ability,while adding a small target dedicated detection head to improve recognition ability in large scenes and small target environments.The experimental results show that the improved model improves accuracy by 0.7％and mAP by 2.1％,reaching 94.5％,thereby achieving real-time detection of abnormal behavior,especially small targets,in power construction sites.
关键词
Yolov7算法/行为检测/小目标检测/注意力机制Key words
Yolov7 algorithm/behavior detection/small object detection/attention mechanism引用本文复制引用
出版年
2024
NETL
NSTL
万方数据