Road’s State Identification based on Residual Network and Convolution Neural Network

In developing countries, road condition usually deteriorate over a period of time, it is a challenge for road conditioning authority to identify and rectify road’s a bad condition such as potholes. Generally, bad road condition arises in the rainy season due to which the transportation domain suffers tremendously.Also today, in the era of automation where the world is looking after reducing manual intervention at minimum, need to have technology through which automatically alert the road maintaining authorities. There are many caveats of road's improper conditions. These include unexpected delays due to traffic, wear and tear of vehicles and affects human health due to unexpected shocks and accidents. Hence, the primary goal of this work is to identify potholes through the help of images that are taken over a period of time from cameras that are installed on roads. This will help road rectification authority to check and identify road’s state by analyzing and processing these images through convolution neural network-based deep learning technique.This paper concentrates on the use of artificial intelligence and concepts of machine learning to automatically identify the conditions of roads and to alert the authorities about it in no time. Data sets were used to train the model about the identification of road quality and accordingly it was tested. For the research work, a number of earlier research papers and deeply evaluated the concepts of Convolution neural network and ResNet has gone through. The paper provides a complete solution to road state detection.