Tung Nguyen

Decode with make_batch_reader

4

Hi team, I have created spark parquet with the schema (and without 'materialize_dataset') below: ``` TrainSchema = Unischema('TrainSchema', [ UnischemaField('features', np.uint8, (DEFAULT_IMAGE_SIZE[0], DEFAULT_IMAGE_SIZE[1], 3), NdarrayCodec(), False), UnischemaField('masks', np.uint8, (DEFAULT_IMAGE_SIZE[0], DEFAULT_IMAGE_SIZE[1]),...