Detection of Potholes in Roads Using Siamese Neural Network for Safe Transportation

In order to develop a safe, secure, and smart transportation system, massive research is being conducted and various steps are being taken by the government and different organizations. But prior detection of potholes in roads is still a major concern as the presence of potholes on roads can lead to sudden and serious accidents and damage to vehicles. Many researchers have proposed different machine learning and deep learning approaches for potholes identification and detection on roads. In this work, Siamese neural network is used for the same as it is more robust to class imbalance and works well with small dataset. In the proposed work, Siamese neural network achieved 99.52% training accuracy and 0.0075 training loss and 99.59% validation accuracy and 0.0108 validation loss. In this paper, first there is a brief introduction presenting the motivation behind the work, followed by a literature review highlighting the key points, limitations, and future scope of the existing research. Second, there is discussion about the dataset and methodologies used followed by a detailed description of the proposed scheme. Last in the results section, the results are shown in the form of graphs and images that have been obtained from the implementation.