Smart Implementation of Computer Vision and Machine Learning for Pothole Detection

As the world is modernizing in each and every aspect on a day to day basis and getting prosperous, so the people are. By this growth in one's individual life in the society is getting on better living standards, under which personal vehicle is emerging as a basic need for everyone to go about in daily life, so that number of vehicles is increasing but road conditions are of mostly the same state or say even the worse keeping all this scenario in the mind we have designed a model to detect and predict the potholes and different anomalies present in the road using different suitable algorithms of machine learning and deep learning. For the execution of the project first, we collected the different road image samples and applied various computer vision operations such as preprocessing steps (different blur, smoothing), morphological operations, canny edge detection and decision tree to detect the pothole. After we have the road images with potholes detected, we made a dataset out of it extracting feature and giving it to the deep neural network (DNN) to process further, and it predicts the potholes using linear regression at the end of the process.