Augmentation
- Right before training start, or before dataset export you will be proposed to augment your dataset. Because 2 images of slightly different colors, or mirrored vertically are completely different to a model, it can increase number of images in your dataset, and, therefore, its quality.
There are 2 types of augmentation available in MakeML. It's image transformations, such as rotations and flips, and color modifications of images, such as change of contrast and brightness.