This is an essential step for a DL pipeline. The existing datasets on pothole images and the solid waste images used in the use cases are pre-processed and are ready to be used for training, validation, and testing. As shown in the following diagram, both the original and modified (additional images downloaded for the pothole class) are organized as sub-folders, each named after one of the five categories and containing only images from that category. There are a few issues to be noted during the training image set preparation:
- Data size: We need to collect at least a hundred images for each class to train a model that works well. The more we can gather, the better the accuracy of the trained model is likely to be. Each of the five categories in the used dataset has more than 1,000 sample images. We also made sure that the images are a good representation...