A Novel Approach for Monitoring Road Way Surface Using CNN Classifier

Abstract

The potholes always refer to a damaged road surface that causes several minor and major road accidents often. These accidents could be avoided by detecting the potholes in prior and taking necessary measures to prevent the accident. To achieve this, we require accurate detection model. So, the model is created using convolutional neural network (CNN) which will be used to classify the conditions of the (normal/damaged) road in real time. The proposed work has a software application that will take an image of road and sent it to a backend flask server and the image is being processed using a developed neural network model and alerts the user to take necessary steps to avoid the accident. The developed model is trained with a dataset that has 300 different images of both normal and pothole roads and based on the classification results it alerts the user by blowing the lights which will be controlled using Atmega Arduino UNO.