Research on Small Object Detection Algorithm in Complex Construction Environment Based on Deformable Convolutional Operator

Industrial safety and construction site management have always been a major concern. The correct wearing of helmet is one of the key measures to reduce industrial safety accidents. This study proposes a method for recognising the correct wearing of safety helmets based on the improved YOLOv5. By analysing the actual detection of the correct wearing of safety helmets by construction workers at actual site locations, the YOLOv5 model is optimised by improving the YOLOv5 model by modifying the convolution operator module and the loss function module in order to further improve the model's detection accuracy for small items. The goal of this study is to achieve automatic monitoring and ensure that staff wear helmets correctly on construction sites and in industrial environments.