Right now the only pre-processing/augmentation is scaling images to [-1, 1]. A good next step would be to add the pre-processing defined in `model/embeddings/inception_preprocessing.py` to the data pipeline.
Right now the only pre-processing/augmentation is scaling images to [-1, 1].
A good next step would be to add the pre-processing defined in
model/embeddings/inception_preprocessing.pyto the data pipeline.