Electric tower defect detection method and system based on Yolo V5

• Main Authors: ZHU ZHU, ZHANG XIAOHANG, HE MAOLI, ZOU WENKAI, ZHANG JIELAN, ZHUANG ZHENHUA
• Format: Patent
• Language: Chinese; English
• Published: 24.06.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention discloses a Yolo V5-based electric tower defect detection method and system. The method comprises the following steps: acquiring a line inspection image; performing Otsu segmentation on the line inspection image to obtain an adaptive threshold T; converting the line inspection image from RGB to HSL; adjusting the brightness of the L channel based on an adaptive threshold T; converting the line inspection image after brightness adjustment from HSL to RGB to complete image adjustment; and inputting the line inspection image after image adjustment into the trained Yolo v5 algorithm model, and carrying out a detection task. Brightness adjustment is carried out based on an adaptive threshold T, namely, when the brightness is smaller than T, a backlight area is represented, and the minimum channel value is increased; when the brightness is greater than T, representing a non-backlight area, and reducing the maximum channel value; after brightness adjustment, image enhancement is carried out, and a subs