Safety Helmet Wearing Detection Method By YOLOv5

In recent years,Accidents caused by not wearing safety helmets have emerged one after another.How to effectively monitor the wearing of safety helmets by workers is an important issue in the field of construction safety.Traditional helmet detection algorithms rely on manually designed features,are susceptible to environmental factors,have poor stability,and have poor detection results.In order to improve the detection effect of safety helmets,this article conducts research from the aspects of feature selection,feature fusion,and network structure improvement.Therefore,this article proposes a safety helmet detection method based on residual network fusion attention mechanism and an improved network structure safety helmet detection method.The experimental results verify the effectiveness of the proposed method.The main work of this article is as follows:For YOLOv5 algorithm in performing small target detection,there are problems such as weak anti-interference ability and many missed detections,this paper proposes a helmet detection method based on residual network fusion attention mechanism.The method first uses residuals to connect deep-shallow feature information to enhance the mutual sharing of information between network and network,then uses attention mechanism to improve the accuracy of the model,and finally replaces the loss function of the original model with EIOU so that the network can better reflect the degree of overlap between objects,thus improving the network detection capability.Experiments are conducted to verify the effectiveness of the improvement ground,and ablation experiments and comparison experiments are also used to verify the improvement effect of the network improvement module and the overall effect of the improved network.To address the problems of poor feature extraction ability and low accuracy of YOLOv5 algorithm for small target detection,this paper proposes a helmet detection method based on improved network structure.The method firstly employs Mascio-9 data enhancement to enrich the target background,secondly introduces the SE-Net attention mechanism to make the model more focused on features that are more critical to the recognition and classification tasks,replaces the PANet layer with the Bi FPN layer to further strengthen the feature extraction capability of the network,introduces the weighted border fusion method instead of the non-maximum suppression method to further improve the missed detection Finally,the loss function is optimized.Relevant experiments are conducted on the homemade dataset,and the experimental results show that the method can efficiently perform helmet detection and further improve the network accuracy compared with the original network model,with an experimental accuracy of 91.76%.The helmet detection algorithm based on the improved YOLOv5 is proposed in this paper.To a certain extent,it can effectively solve the problems of weak anti-interference ability of the model,low accuracy and more leakage detection,and there is a certain application space in the development of the construction production safety field.