Coal gangue target detection method based on improved YOLOv5 algorithm

In order to solve the problem that an existing coal and gangue recognition algorithm is unstable, the invention provides a coal and gangue target detection method based on an improved YOLOv5 algorithm, and depth separable convolution is introduced into a backbone network to reduce the number of original network model parameters so as to improve the detection speed of the network; a convolution block attention model is introduced to enhance the saliency of a coal and gangue target in an image, and the problem that the coal and gangue target on a conveying belt is significantly reduced due to factors such as illumination, and consequently the target is difficult to accurately detect is solved. In order to solve the problem that a coal gangue small target is difficult to detect, a detection layer is added to a head part of an original network, and multi-scale detection of the head part is achieved; finally, the improved YOLOv5 algorithm is trained, and a final detection network is obtained.The coal and gangue ta