Scholarlink

Safety helmets are crucial for protecting workers in hazardous construction environments. The core idea of YOLO is to transform object detection into a regression problem, using the entire image as input to the network and only passing through a neural network to obtain the position and category of the bounding box. Our approach integrates GhostNetv2, attention modules (SCNet, CANet), and the GAM optimizer to boost accuracy and generalization. Experimental results confirm that these enhancements significantly improve efficiency, adaptability, and detection performance in real-world conditions. This study enhances YOLO-based helmet detection by improving mAP by 2% while reducing parameters and FLOPs by over 25%.

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

Better YOLO with Attention-Augmented Network and Enhanced Generalization Performance for Object Detection

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Shuqi Shen,
  • Haoting Liu

摘要

Safety helmets are crucial for protecting workers in hazardous construction environments. The core idea of YOLO is to transform object detection into a regression problem, using the entire image as input to the network and only passing through a neural network to obtain the position and category of the bounding box. Our approach integrates GhostNetv2, attention modules (SCNet, CANet), and the GAM optimizer to boost accuracy and generalization. Experimental results confirm that these enhancements significantly improve efficiency, adaptability, and detection performance in real-world conditions. This study enhances YOLO-based helmet detection by improving mAP by 2% while reducing parameters and FLOPs by over 25%.