An Advanced Deep Learning Approach for Safety Helmet Wearing Detection

Abstract

Automatically detect whether workers are wearing safety helmet at construction site is significant for safety production. Concerning the problem that the existing safety helmet wearing detection method is difficult to detect the partial occlusion, different size and small object, and the detection accuracy is low. In this paper, we present an advanced deep learning based approach to determine whether workers are wearing safety helmets. In our framework, we first use the multi-scale training and the increasing anchors strategies to enhance the robustness of the original Faster RCNN algorithm to detect different scales and small object. Then, the Online Hard Example Mining (OHEM) is to optimize model to prevent the imbalance of positive and negative samples. Finally, the person wearing the helmet and its parts (helmet and person) are detected by improved Faster RCNN, the multi-part combination method uses the geometric information of the detection objects to determine if a worker is wearing a helmet. Experiments show that compared with the original Faster RCNN, the detection accuracy is increased by 7%.