DCMS-YOLOv5: A Dual-Channel and Multi-Scale Vertical Expansion Helmet Detection Model Based on YOLOv5

this paper proposes Dual Channel Multi Scale YOLOv5 (DCMS-YOLOv5), an improvement of YOLOv5, to increase the recognition accuracy of helmet detection methods. The model has a dual-channel architecture, and feature extraction and fusion are performed in a lateral connection to enhance the model's ability to capture targets in complex scenes. The features and local dependencies are characterized at multiple scales to improve the model's ability to capture small targets. The model is validated with the safety helmet-wearing dataset (SHWD) and compared with other methods. The experimental results show that the DCMS-YOLOv5 model provides high helmet detection accuracy, performs excellent for detecting small targets, and has strong generalization ability.