Smoking Detection Algorithm Based On Improved YOLOv5
Qingyan Ding, Xuecheng Dong, Weimeng Guo, Zheng Wan, Yu Rong Pan · 2023
Aiming at the problem of low detection accuracy of small and medium-sized targets in complex scenes, a smoking detection method in public places based on improved YOLOv5 was proposed.Firstly, Mosaic-9 was used to replace the traditional Mosaic data enhancement algorithm to increase the amount of image mixing and improve the learning ability of the algorithm model for small targets. Secondly, the dual channel attention mechanism (CBAM) is introduced to pay more attention to the location of the target in the image and improve the recognition ability of the algorithm model. Then, the upsampling method is changed to transpose convolution, so that the model can better recognize the semantic information in the image and improve the recognition ability of the target region. Then, the SPD-Conv module is introduced to improve the recognition ability of low resolution image and small target image.