Transmission line insulator defect detection algorithm based on MAP-YOLOv8

As society's expectations for the stability and quality of power supply increase, it has become critical to ensure the safe and stable operation of transmission lines. Insulators play a crucial role in the design of transmission lines and are mainly responsible for fixing conductors and providing conductor insulation. Therefore, real-time and accurate monitoring of the status of insulators on transmission lines has become a top priority in power inspection. To this end, this paper proposes a transmission line insulator defect detection algorithm based on MAP-YOLOv8. Firstly, the network structure model of YOLOv8 is improved, by adding the GSConv module to the neck layer and introducing the SimSPPF module to optimize the feature pyramid layer, which reduces the complexity of the network computation and improves the model's computational efficiency at the same time; secondly, this paper proposes a new attention mechanism (MAP-CA), which, by combining the mean pooling and the maximum pooling, the mechanism effectively fuses global and local information to achieve higher recognition accuracy; finally, by super-resolution reconstruction and enhancement of insulator images, it solves the problems of poor image quality and loss of detail information due to low image pixel resolution. The experimental results show that the average accuracy of MAP-YOLOv8 reaches 96.6%, which is 14% higher than that of the YOLOv8 base model, with a memory usage of 8.3 MB and an F1 score of 0.981. A detection speed of 89 frames per second is achieved, which meets the requirement of real-time detection of insulator defects.