基于YOLO的视频行人检测研究

Research on Video Pedestrian Detection Based on YOLO

针对视频中行人检测准确率低的问题,提出一种基于YOLO的视频行人检测研究方法。首先引入YOLOv5检测器,在YOLOv5的Neck部分融合注意力模块CBAM,加强对低层特征的提取,解决视频中行人运动模糊问题,提高行人检测精度;其次引入DeepSort算法,在视频行人数据集上进行训练,实现行人跟踪;最后在DeepSort算法实现行人跟踪后引入REID技术,有效纠正行人运动轨迹,解决行人位置信息出错问题。实验结果表明:所提方法较原始算法mAP@0.5提高了2.8%,mAP@0.5:0.95提高了5.4%。

Aiming at the low accuracy of pedestrian detection in video,a research method of video pedestrian detection based on YOLO was proposed.Firstly,YOLOv5 detector is introduced,and attention module CBAM is integrated into THE Neck part of YOLOv5 to enhance the extraction of low-level features,solve the problem of blurred pedestrian movement in the video,and improve pedestrian detection accuracy.Secondly,DeepSort algorithm is introduced to train on video pedestrian data set to realize pedestrian tracking.Finally,REID technology is introduced after DeepSort algorithm realizes pedestrian tracking to effectively correct pedestrian movement trajectory and solve the error problem of pedestrian location information.The experimental results show that the proposed method improves by 2.8%and 5.4%compared with the original method mAP@0.5 and mAP@0.5:0.95.