Aircraft sheet metal riveting practical training safety protection detection algorithm based on YOLOv5

In the field of aircraft sheet metal riveting practical training safety and protection monitoring, it is crucial to accurately detect whether students are wearing goggles and noise-proof headphones. In this study, a dataset containing 4650 images was constructed for training and evaluating a YOLOv5-based target detection model. The YOLOv5 network structure is improved by optimizing the YOLOv5 network structure, replacing the backbone feature extraction network with PLCNet, using CNeB module for the Neck layer, and replacing GSConv with ordinary convolution. The experimental results show that the full-category mAP is significantly increased from 85.3% to 91.5%, which effectively enhances the model's ability to detect students' wearing of protective equipment in practical training, provides powerful technical support for improving the safety management level of aircraft sheet metal riveting practical training, and helps to create an intelligent and efficient safety monitoring system for aircraft maintenance practical training.