I want to create an image dataset of leaf dataset. For this i captured images and found 6 classes, each classes has roughly 70-60 images. Noe while creating training and validation set, i know i shouldn't augment the validation set. in that case if at first I separate the training and validation set (80%-20%) then in validation set i will have roughly 15-20 images per class. and after augmentation in training set there shall be 500-600 images per class (or depends on my augmentation). Now my question, such low number of images in validation set will be a problem? if it is then what should I do to solve it?
I tried augmenting using tensorflow.
I suggest not applying data augmentation on the test set/validation set as variation and diversity in the dataset is required for training the model and the train/validation set is just used to check the performance of the trained model.