Intelligent identification of miners'safety in coal mines is one of the important protection measures to prevent miners from accidental injury.In order to improve the recognition accuracy in the dark environment such as insufficient light in the coal mine,a target detection algorithm based on improved YOLOv5 was proposed to intelligently identify the safe wearing of miners.Firstly,the data is collected in the field to construct a secure wearable data set,which is input into the weak light enhanced network Zero-DCE to improve the generalization ability of the model.Secondly,the C-ASPP module is proposed.By improving the ASPP and adding the attention mechanism,it is added to the backbone network to make the model pay more attention to the characteristics of the safe wearing area.Then,the Transformer algorithm is integrated into the backbone to enhance the dynamic adjustment ability of the model to different scale targets.Finally,in the feature fusion stage,the bidirectional feature fusion pyramid model is used to improve the feature extraction ability and detection performance of the model.The results show that the average detection accuracy of the improved YOLOv5 algorithm is increased to 90.2％,which is 3 percentage points higher than that of the original algorithm,and the detection speed is 81.2 frames/s.Compared with other algorithms,it has higher accuracy and speed,which can meet the requirements of miners'safe wearing recognition in the underground working area.
关键词
安全智能识别/改进YOLOv5算法/井下暗环境/矿工安全穿戴
Key words
Safety intelligent identification/Improved YOLOv5 algorithm/Underground dark environment/Safe wearing of miners
维普
万方数据