"Fix My Way" Using Yolov8 and Machine Learning

Abstract

Roadway infrastructure forms the backbone of the social and economic growth of an area, but potholes adversely affect road condition, causing accidents and damage to vehicles. As a solution to the problem, we introduce "Fix My Way," a web application based on machine learning to identify and authenticate potholes on roads in Madhya Pradesh, India. The front end, developed using HTML, CSS, and JavaScript, enables the user to provide complaints with images, location, contact information, and landmarks. Images are scanned through the YOLOv8 model, It is  labeled Kaggle dataset through Roboflow, to identify potholes. Only validated complaints are inserted into a PostgreSQL database. Interaction between the front end, ML model, and database is achieved through a backend using Flask. The system contains a complaint page to ensure transparency and enable tracking of submissions by users. Through image validation and geolocation enabling effective, priority-based road maintenance and intelligent infrastructure monitoring.

Keywords: Pothole Detection YOLOv8 Object Detection, Full-Stack Development, Flask API, PostgreSQL, Road Infrastructure, Highway Complaint System