Research Article
An Improved Detection Method of Safety Helmet Wearing Based on CenterNet
@INPROCEEDINGS{10.1007/978-3-030-82562-1_20, author={Bo Wang and Yong Zhang and Qinjun Zhao and Shengjun Shi}, title={An Improved Detection Method of Safety Helmet Wearing Based on CenterNet}, proceedings={Multimedia Technology and Enhanced Learning. Third EAI International Conference, ICMTEL 2021, Virtual Event, April 8--9, 2021, Proceedings, Part I}, proceedings_a={ICMTEL}, year={2021}, month={7}, keywords={Improved detection Safety helmet CenterNet}, doi={10.1007/978-3-030-82562-1_20} }- Bo Wang
Yong Zhang
Qinjun Zhao
Shengjun Shi
Year: 2021
An Improved Detection Method of Safety Helmet Wearing Based on CenterNet
ICMTEL
Springer
DOI: 10.1007/978-3-030-82562-1_20
Abstract
In some factories or construction sites, accidents occur because workers do not wearing safety helmets correctly. In order to reduce the accident rate, an improved detection method of safety helmet wearing based on CenterNet algorithm is proposed. The original IOU method is optimized by combining with GIoU, and debug the training model Res/DLA framework in the training process. At the same time, various parameters are adjusted by experiments. In the safety helmet wearing test task, theoretical analysis and experimental results show that mAP (Mean Average Precision) is up to 42.6%, detection rate is increased to 30.3%. Compared with CenterNet, the detection accuracy and detection rate are slightly improved. The proposed algorithm not only meets the real-time performance of detection task in safety helmet wearing detection but also has higher detection accuracy.
