CNN-MAO: Convolutional Neural Network-based Modified Aquilla Optimization Algorithm for Pothole Identification from Thermal Images

Potholes are the most common cause of accidents on the road surface, and the primary cause is water. Potholes in the pavement can be formed by a number of things, including gasoline or fuel leaks, automobile smearing, and the disposal of rock cuttings. As a result, in order to avoid accidents, it is essential to identify the pothole in advance, either automatically or manually. There are several ways to detect the pothole manually, nonetheless, they consume more time, power, and high setup cost. However, the automatic detection follows an optical imaging system, and the detection of pothole during bad weather conditions and night-time become arduous. Hence we proposed a novel method to detect the pothole by using a thermal imaging system known as convolutional neural network (CNN)-based modified aquilla optimization (AO) algorithm. The proposed method follows Data acquisition, Image preprocessing, and Data augmentation processes prior to the application of classification tasks. The proposed CNN-based Modified AO approach enhances the classification accuracy, precision, recall, and F1-score. However, it minimizes the classification error and detection time. The performances of our proposed work are compared with other approaches such as CNN, CNN-TI, YOLO-NN, and DNN. The experimental analysis also depicts that our proposed work has better performances than the other approaches.