Application of Thermal Imaging Based on Improved Faster R-CNN Algorithm in Distribution Cabinet Maintenance

Due to the influence of environment, light, shooting angle and other factors, the infrared image will have different degrees of distortion, fuzzy distortion or noise mixing, which will cause the degradation of image quality. This paper proposes an application of thermal imaging based on improved Faster R-CNN algorithm in the maintenance of power distribution cabinet. Firstly, the collected infrared image of the distribution cabinet is pre-processed by graying and background subtraction; Secondly, the idea of non-maximum suppression (NMS) is introduced to improve the Fast RCNN, and the category correction function is added, so that the model can correct the recognition error of equipment part type. The experiment shows that the accuracy, recall and F1 values of this method are 0.895, 0.784 and 0.836, respectively. Therefore, the proposed method has good practicability for infrared thermal imaging detection of electrical equipment.