| images = [ | |
| train_data_augmentation(convert_to_tf_tensor(image.convert("RGB"))) for image in example_batch["image"] | |
| ] | |
| example_batch["pixel_values"] = [tf.transpose(tf.squeeze(image)) for image in images] | |
| return example_batch | |
| def preprocess_val(example_batch): | |
| """Apply val_transforms across a batch.""" |