Pothole Detection Model based on Convolutional Neural Network and Metaheuristic Group Learning Algorithm
Main Article Content
Abstract
In this paper, we have presented a pothole detection model in order to enhance the road safety. In this model, convolutional neural network (CNN) and metaheuristic group learning algorithm (CNN) is employed for classify the pothole. The proposed model has four stages: reading the dataset, pre-processing the dataset, classification of potholes using machine learning, and performance evaluation. In the first stage, a standard dataset is read. In the second stage, the pre-processing of the dataset is done using the filter, enhancement, and segmentation methods to find the region of interest. Further, in this stage, a metaheuristic group learning algorithm is utilized to fine-tune the pre-processing method. Next, in the third stage, the classification of potholes is done using CNN algorithms. To accomplish this goal, the CNN algorithm is trained and tested using the dataset generated after the pre-processing method. Finally, the evaluation of the proposed model is done using various performance metrics such as accuracy, precision, recall, and F1-score. The evaluation of the proposed model is done on the standard dataset of potholes, which is available on the Kaggle database. This dataset contains smooth and pothole images. The result shows that the proposed model achieves better performance metrics than the existing method.