Research on Insulator Detection Method Based on Scene Recognition

Target recognition of insulators is the prerequisite for the condition assessment of insulator equipment. Accurate identification of insulators is of great significance to insulator maintenance. This paper combines infrared images and machine learning to propose an infrared image insulator detection method for scene recognition including image preprocessing, model prediction and image fusion; construct a target detection model through the structure of encoding and decoding in semantic segmentation, which can accurately identify insulators. The accuracy of the insulator detection method based on scene recognition proposed in this paper has reached 99.6%, while the traditional target recognition method is 44%. It provides solutions for field applications in the field of embedded devices and intelligent robots.