[Python] Low validation accuracy during data training using lfw datasets

So i was training 400 labels of image datasets that consist around 900 images that split into 80% training and 20% validation. i'm following a guide from tensorflow here(https://www.tensorflow.org/tutorials/images/transfer_learning_with_hub)

this is my dataset (https://drive.google.com/drive/folders/1yIEig6K3g3Y2gFudkE0ca64UzkQtsORA?usp=drive_link)

this is preprocessed dataset using MTCNN

should i change my dataset or else?

Continue reading...