arXiv 2307.041032023-07-09cs.CV

CA-CentripetalNet：一种用于安全帽佩戴检测的新型无锚点深度学习框架

CA-CentripetalNet: A novel anchor-free deep learning framework for hardhat wearing detection

School of Information Engineering, Guangdong University of Technology（广东工业大学信息工程学院）；State Key Laboratory of Precision Electronic Manufacturing Technology and Equipment, Guangdong University of Technology（广东工业大学精密电子制造技术与设备国家重点实验室）；School of Mechanical and Electrical Engineering, Guangdong University of Technology（广东工业大学机电工程学院）

Zhijian Liu, Nian Cai, Wensheng Ouyang, Chengbin Zhang, Nili Tian, Han Wang

备注：: It has been accepted for the journal of Signal, Image and Video Processing, which is a complete version. It is noted that it has been deleted for future publishing
发表信息：: Signal, Image and Video Processing,2023

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出无锚点深度学习框架CA-CentripetalNet，通过垂直-水平角点池化和边界约束中心注意力提升特征提取与利用能力，在安全帽佩戴检测中取得86.63% mAP，尤其在小目标和未佩戴场景下优于现有方法。

摘要：

自动安全帽佩戴检测可以加强建筑工地的安全管理，但由于视频监控场景复杂，这一任务仍然具有挑战性。针对以往基于深度学习的方法泛化能力较差的问题，本文提出了一种新型无锚点深度学习框架CA-CentripetalNet，用于安全帽佩戴检测。为了提高CA-CentripetalNet的特征提取和利用能力，提出了两种新方案：垂直-水平角点池化和边界约束中心注意力。前者旨在实现边缘特征和内部特征的综合利用；后者旨在强制骨干网络关注内部特征，且仅在训练阶段使用，而不在检测阶段使用。实验结果表明，与现有的基于深度学习的方法相比，CA-CentripetalNet在合理的速度下以更少的内存消耗取得了更好的性能，mAP（平均精度均值）达到86.63%，尤其是在小尺寸安全帽和未佩戴安全帽的情况下表现更优。

英文摘要：

Automatic hardhat wearing detection can strengthen the safety management in construction sites, which is still challenging due to complicated video surveillance scenes. To deal with the poor generalization of previous deep learning based methods, a novel anchor-free deep learning framework called CA-CentripetalNet is proposed for hardhat wearing detection. Two novel schemes are proposed to improve the feature extraction and utilization ability of CA-CentripetalNet, which are vertical-horizontal corner pooling and bounding constrained center attention. The former is designed to realize the comprehensive utilization of marginal features and internal features. The latter is designed to enforce the backbone to pay attention to internal features, which is only used during the training rather than during the detection. Experimental results indicate that the CA-CentripetalNet achieves better performance with the 86.63% mAP (mean Average Precision) with less memory consumption at a reasonable speed than the existing deep learning based methods, especially in case of small-scale hardhats and non-worn-hardhats.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。