Ensemble-based approach using inception V2, VGG-16, and Xception convolutional neural networks for surface cracks detection

: Manual road crack detection is time-consuming. However, deep learning-based solutionsare quick and accurate. Various deep learning-based convolutional neural networks (CNN) have beenrecently proposed. This study implies a comprehensive assessment of the performance of inceptionV2, VGG16, and Xception CNN utilizing the surface cracks dataset. The research approach comprisesfour distinct steps. Training and validating these pre-trained models are necessary by immobilizingcertain foundational layers. The previously frozen layers are thawed during the second stage, and thetraining and validation process is repeated. Subsequently, the performance of the model is evaluated.To enhance the performance of the models in detecting surface cracks in dataset images, aftercompletion of the model training and validation process for both frozen and unfrozen layers, themodels are combined using the ensemble technique to increase the overall performance for surfacecrack detection. The performance of the models, including inception V2, VGG16, Xception, and theensemble model, is evaluated using evaluation metrics including accuracy, precision, recall, and F1score. The ensemble has the highest precision 99.97% and the highest recall 99.92%. along with thehighest accuracy 99.93% and F1 score 99.92%, compared to the other CNN models.