Hi!
I’d like to know if there’s a way to shuffle annotated images before i export them on training/test/valid datasets.
Classes are shown to be leaned to one side.
Thanks in advance. 
Hi @ljy9969 you can change your train/valid/test split in Step 2 of dataset version generation:
Select “Rebalance” and choose the split of your choice. Note that if you add augmentations, the split will change, as only training set images are augmented, whereas preprocessing applies to all images.
That did not answer the question, as far as I understood it. I, too, need to shuffle the data to ensure that validation and testing are representative of the training data. This I suppose should be done before the rebalancing into train/val/test split.