Lightweight defect detection model for power grid equipment based on dual branch fusion network

Given the insufficient detection accuracy of existing models, this study proposes a lightweight defect detection model for power grid equipment based on a dual branch fusion network. Based on a dual branch fusion network, an enhanced feature cascade method is adopted to extract local and global features through a dual branch structure, and an adaptive fusion mechanism is used to achieve effective feature integration. Classify different types of equipment defects into different categories, and the detection head detects defects in power grid equipment by outputting the position information and defect category information of the prediction box. The test results show that the average accuracy of the network model remains stable, the parameter count remains at a low level, and a good balance is achieved between model complexity and computational efficiency; When detecting insulators and their defects, it is possible to promptly and accurately identify the location of the defects. The dual improvement of detection accuracy and model lightweighting has been achieved, ensuring the safe operation of power grid equipment.