Road Surface Analysis Through Prognostic Model Using Image Processing and Deep Learning Techniques

Road surface quality and condition become crucial factors for driving ease and safety. Sophisticated tools are available for tackling various problems in the reconstruction and study of road surfaces. For maintaining road infrastructure and assuring the safety, accurate road surface distress detection, analysis, and categorization are essential. Road surface distress detection and classification using conventional approaches can be labor-intensive, cumbersome, and prone to mistakes. The goal of this study is to automatically detect and classify the types of road surface distress like potholes and cracks using Image Processing and Deep Learning Techniques. Recurrent Neural Networks (RNN) with Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) are used to evaluate the classification of road surface distress. With an accuracy of 89.1%, it has been discovered that the Recurrent Neural Network outperforms the other classifier. Through its ability to enhance road safety, reduce maintenance costs, and improve the driving experience, this advancement has the potential to significantly impact society.