Scholarlink

<p>To address challenges in detecting railway workers, safety helmets, and reflective vests with existing algorithms, particularly in open and complex railway environments with high missed detection rates, we designed and implemented an improved real-time detection algorithm based on YOLOv5. This algorithm aims to detect workers in railway environments and ensure that they are wearing safety helmets and reflective vests correctly. Specifically, we improved the YOLOv5 algorithm by incorporating the CBAM into all C3 modules within the bottleneck to improve the model’s feature extraction capabilities. Additionally, we replaced the original CIoU loss function with the MPDIoU loss function to address varying aspect ratios, which is crucial for detecting safety helmets and reflective vests of different dimensions. To confirm the feasibility of the proposed algorithm, we conducted comparative experiments using a railroad worker safety detection dataset with various improved algorithms and performed ablation studies to evaluate the contribution of each improvement. The improved algorithm achieved a mAP50-95 of 70.9% and a recall of 93.4%, 8.2% and 8.1% greater than those of the original YOLOv5 algorithm, respectively. The improved algorithm significantly reduces the missed detection rate in complex environments, dense scenes, and small-target scenarios, ensuring the safety of railway workers. This study is open source, and the project can be accessed at <a href="https://github.com/112345434/yolov5-master">https://github.com/112345434/yolov5-master</a>.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Safety equipment compliance analysis for occupational safety

  • 期刊论文

  • 发表日期：

    2025年6月10日

查看全文
  • Yachao Si,
  • Jiajie Gao,
  • Xingxuan Zhao,
  • Shaojie Li

摘要

To address challenges in detecting railway workers, safety helmets, and reflective vests with existing algorithms, particularly in open and complex railway environments with high missed detection rates, we designed and implemented an improved real-time detection algorithm based on YOLOv5. This algorithm aims to detect workers in railway environments and ensure that they are wearing safety helmets and reflective vests correctly. Specifically, we improved the YOLOv5 algorithm by incorporating the CBAM into all C3 modules within the bottleneck to improve the model’s feature extraction capabilities. Additionally, we replaced the original CIoU loss function with the MPDIoU loss function to address varying aspect ratios, which is crucial for detecting safety helmets and reflective vests of different dimensions. To confirm the feasibility of the proposed algorithm, we conducted comparative experiments using a railroad worker safety detection dataset with various improved algorithms and performed ablation studies to evaluate the contribution of each improvement. The improved algorithm achieved a mAP50-95 of 70.9% and a recall of 93.4%, 8.2% and 8.1% greater than those of the original YOLOv5 algorithm, respectively. The improved algorithm significantly reduces the missed detection rate in complex environments, dense scenes, and small-target scenarios, ensuring the safety of railway workers. This study is open source, and the project can be accessed at https://github.com/112345434/yolov5-master.