In order to identify the closer damper,a damper automatic detection method based on improved YOLOv4 was proposed.In YOLOv4 method,non maximum suppression method with fixed threshold was used to select the detection frame,lower threshold would lead to the loss of highly overlapped targets,but higher threshold would lead to more false detection.Therefore,a dynamic non maximum suppression method was proposed and was used to YOLOv4 target detection.By using the method,the dynamic threshold was determined according to the statistical characteristics of the detection frame around the target,the accuracy of boundary box selection was improved,and the probability of false detection and missing detection in the detection of high overlap damper was reduced.In order to further improve the detection accuracy of damper,the piecewise linear function was used as the activation function to overcome the problem that Leaky ReLU function in YOLOv4 algorithm did not deal with negative value very well and the function curve was not smooth,the nonlinear expression ability of the model was enhanced.The results show that by using the improved YOLOv4 damper target detection method,the overlapping dampers can be detected well,and it has higher detection accuracy.