POWERLINE DEFECT DETECTION USING COMPUTER VISION MODEL

ABSTRACT

Powerlines, essential for the functioning of modern society, face numerous challenges, with insulator defects being a significant concern. This project focuses on developing a robust detection system using advanced Computer Vision models to identify insulator defects accurately and efficiently. By leveraging a model called YOLO v8, which is an important model for object detection, this study seeks to provide a vast application in the aspect of powerlines defect detection. Through rigorous analysis of visual data and the creation of sophisticated algorithms, the goal is to provide infrastructure managers with the means to detect insulator defects early, preventing potential catastrophic failures. Upon completion of the project a software solution capable of receiving images in popular formats such as JPEG and PNG was successfully developed. The software undergoes image processing to identify any defects present, effectively communicating their nature and location to the user. The software achieved an accuracy rate of 71.2% and a precision rate of 55.6%. If implemented, this technology could serve as a foundational element in a broader infrastructure, enhancing the dependability of power distribution not only domestically but also internationally.