Chinese Journal of Liquid Crystals and Displays, Volume. 40, Issue 3, 505(2025)
Dense pedestrian detection algorithm based on YOLOv7 with optimized weights
Figures & Tables(10)
Fig. 6. Comparison between the YOLOv7 algorithm （a） and the algorithm in this paper （b）
Table 1. Ablation experiment results on Wider-Person dataset
View tableView in ArticleTable 1. Ablation experiment results on Wider-Person dataset
算法 评价指标 P R mAP Params/M YOLOv7 0.805 0.786 0.798 37.19 YOLOv7+EMA 0.811 0.798 0.806 37.19 YOLOv7+SimAM 0.809 0.800 0.805 37.19 YOLOv7+EMA+SimAM 0.816 0.805 0.812 37.19 YOLOv7+ELCM 0.825 0.811 0.816 31.49 YOLOv7+Focal-SIoU 0.816 0.799 0.806 37.19 YOLOv7+ELCM+EMA+Focal-SIoU 0.842 0.815 0.837 31.49
Table 2. Comparison of the effects of SPPCSPC and ELCM modules
View tableView in ArticleTable 2. Comparison of the effects of SPPCSPC and ELCM modules
模块 mAP@0.5 参数量/M 时间/s SPPCSPC 0.798 37.19 270 ELCM 0.816 31.49 220
Table 3. Comparison of different algorithms on the Wider Person dataset
View tableView in ArticleTable 3. Comparison of different algorithms on the Wider Person dataset
算法 评价指标 mAP Params/M SSD 0.574 26.16 Faster-RCNN 0.615 137.08 EfficientNet 0.694 3.87 DETR 0.498 41.17 RetinaNet 0.704 36.78 YOLOv5s 0.698 7.01 YOLOv7 0.798 37.19 YOLOv8s 0.772 11.14 YOLOv9s 0.781 9.74 YOLOv10s 0.768 8.07 本文算法 0.837 31.49
Table 4. Comparison of different algorithms on the Crowd Human dataset
View tableView in ArticleTable 4. Comparison of different algorithms on the Crowd Human dataset
算法 评价指标 mAP Params/M SSD 0.679 26.16 Faster-RCNN 0.802 137.08 EfficientNet 0.726 3.87 DETR 0.504 41.17 RetinaNet 0.721 36.78 YOLOv5s 0.767 7.01 YOLOv7 0.785 37.19 YOLOv8s 0.765 11.14 YOLOv9s 0.772 9.74 YOLOv10s 0.753 8.07 本文算法 0.826 31.49
Tools
Get Citation
Copy Citation Text
Jie CAO, Yu NIU, Haopeng LIANG. Dense pedestrian detection algorithm based on YOLOv7 with optimized weights[J]. Chinese Journal of Liquid Crystals and Displays, 2025, 40(3): 505
Paper Information
Category:
Received: Jun. 17, 2024
Accepted: --
Published Online: Apr. 27, 2025
The Author Email: Jie CAO (haop1115@163.com)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20