Comparison of Tracking Performance Among Different Improvement Methods with YOLOv5s (↓represents “Lower is better”, ↑ represents “Higher is better”).

Comparison of Tracking Performance Among Different Improvement Methods with YOLOv5s (↓represents “Lower is better”, ↑ represents “Higher is better”).

<p>Comparison of Tracking Performance Among Different Improvement Methods with YOLOv5s (↓represents “Lower is better”, ↑ represents “Higher is better”).</p>

Saved in:

Bibliographic Details
Main Author:	Guofeng Qin (11025170) (author)
Other Authors:	Rongting Pan (22560172) (author), Yi Deng (421971) (author), Peiwen Mi (20678488) (author), Yongjian Zhu (6562817) (author)
Published:	2025
Subjects:	Ecology Cancer Space Science Biological Sciences not elsewhere classified Information Systems not elsewhere classified small object scenarios pixel small object experimental results show eiou loss function effectively track targets ciou loss function 9 %, respectively lightweight shufflenetv2 network poor tracking stability model &# 8217 tracking algorithm based road pedestrian detection deepsort tracking framework 5 %, respectively 3 %, improving yolov5s detection network improved yolov5s achieves deepsort </ p pedestrian detection improved yolov5s deepsort achieves 3 %, backbone network original algorithm original yolov5s enhanced yolov5s detection performance detection layer model size model parameters xlink "> traffic roads paper proposes neck structure maintaining accuracy low accuracy identity switches head self dense occlusion
Tags:	Add Tag No Tags, Be the first to tag this record!

Cannot write session to /tmp/vufind_sessions/sess_rlkua8j774346drcg3ocnpltm7