Fig. 7

Illustrates the dataset utilization process. The dataset utilization process involves loading the training and test images into a deep learning model (such as DeepLabV3 + , UNet, SegNet) using Python’s PIL library and relevant modules/packages from the OpenCV library. The model is capable of performing pixel-level semantic segmentation tasks. To evaluate the predictive performance of CNN models, standard metrics such as Accuracy, Precision, Recall, and F1-score are utilized for quantitative prediction assessment.