Infrared Image Segmentation and Temperature Monitoring Based on YOLO Model Object Detection Results

The safety and reliability of power equipment are critical. It requires advanced monitoring and diagnostic techniques. To process infrared images of power equipment, traditional methods have many limitations, including the need for expert analysis and the potential for subjective interpretation. Recent researches focused on overcoming these challenges through automation and enhanced accuracy. This paper introduced a streamlined approach using the YOLO (You Only Look Once) detection model of processing and recognizing for equipment infrared images in power substations. It utilizes the YOLO object detection algorithm to identify components such as bushings and surge arresters. Then, upon completing image segmentation and reconstructing the temperature matrix, it assesses the equipment's risk of overheating. The integration of the YOLO detection model allows for effective recognition and isolation of key equipment against complex backgrounds. Analysis of temperature data from segmented components aids in determining safe thermal operation thresholds. This automated process enhances the accuracy of thermal condition assessments and reduces manual intervention, streamlining maintenance workflows. Findings suggest this method is a reliable tool for early detection of potential overheating, preventing equipment failures and ensuring continuous operation of power equipment.