Hello everybody,
I want to ask you about the example binary segmentation (camvid), how does data augmentation with albumentation work exactly? does it generate a new batch of images each step in the training ?
My dataset is very small, it contains 40 images for training and 10 images for validation. should i use this type of data augmentation or generate more images and save them to a directory and then do the training.
I will be grateful if you can help me or give a hint about how to proceed.
Hello everybody, I want to ask you about the example binary segmentation (camvid), how does data augmentation with albumentation work exactly? does it generate a new batch of images each step in the training ? My dataset is very small, it contains 40 images for training and 10 images for validation. should i use this type of data augmentation or generate more images and save them to a directory and then do the training. I will be grateful if you can help me or give a hint about how to proceed.