transformers/tests/models/donut/test_image_processing_donut.py

# Copyright 2022 HuggingFace Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.


import unittest

import numpy as np

from transformers.testing_utils import is_flaky, require_torch, require_vision
from transformers.utils import is_torch_available, is_torchvision_available, is_vision_available

from ...test_image_processing_common import ImageProcessingTestMixin, prepare_image_inputs


if is_torch_available():
    import torch

if is_vision_available():
    from PIL import Image

    from transformers import DonutImageProcessor

    if is_torchvision_available():
        from transformers import DonutImageProcessorFast


class DonutImageProcessingTester:
    def __init__(
        self,
        parent,
        batch_size=7,
        num_channels=3,
        image_size=18,
        min_resolution=30,
        max_resolution=400,
        do_resize=True,
        size=None,
        do_thumbnail=True,
        do_align_axis=False,
        do_pad=True,
        do_normalize=True,
        image_mean=[0.5, 0.5, 0.5],
        image_std=[0.5, 0.5, 0.5],
    ):
        self.parent = parent
        self.batch_size = batch_size
        self.num_channels = num_channels
        self.image_size = image_size
        self.min_resolution = min_resolution
        self.max_resolution = max_resolution
        self.do_resize = do_resize
        self.size = size if size is not None else {"height": 18, "width": 20}
        self.do_thumbnail = do_thumbnail
        self.do_align_axis = do_align_axis
        self.do_pad = do_pad
        self.do_normalize = do_normalize
        self.image_mean = image_mean
        self.image_std = image_std

    def prepare_image_processor_dict(self):
        return {
            "do_resize": self.do_resize,
            "size": self.size,
            "do_thumbnail": self.do_thumbnail,
            "do_align_long_axis": self.do_align_axis,
            "do_pad": self.do_pad,
            "do_normalize": self.do_normalize,
            "image_mean": self.image_mean,
            "image_std": self.image_std,
        }

    def expected_output_image_shape(self, images):
        return self.num_channels, self.size["height"], self.size["width"]

    def prepare_image_inputs(self, equal_resolution=False, numpify=False, torchify=False):
        return prepare_image_inputs(
            batch_size=self.batch_size,
            num_channels=self.num_channels,
            min_resolution=self.min_resolution,
            max_resolution=self.max_resolution,
            equal_resolution=equal_resolution,
            numpify=numpify,
            torchify=torchify,
        )


@require_torch
@require_vision
class DonutImageProcessingTest(ImageProcessingTestMixin, unittest.TestCase):
    image_processing_class = DonutImageProcessor if is_vision_available() else None
    fast_image_processing_class = DonutImageProcessorFast if is_torchvision_available() else None

    def setUp(self):
        super().setUp()
        self.image_processor_tester = DonutImageProcessingTester(self)

    @property
    def image_processor_dict(self):
        return self.image_processor_tester.prepare_image_processor_dict()

    def test_image_processor_properties(self):
        for image_processing_class in self.image_processor_list:
            image_processing = image_processing_class(**self.image_processor_dict)
            self.assertTrue(hasattr(image_processing, "do_resize"))
            self.assertTrue(hasattr(image_processing, "size"))
            self.assertTrue(hasattr(image_processing, "do_thumbnail"))
            self.assertTrue(hasattr(image_processing, "do_align_long_axis"))
            self.assertTrue(hasattr(image_processing, "do_pad"))
            self.assertTrue(hasattr(image_processing, "do_normalize"))
            self.assertTrue(hasattr(image_processing, "image_mean"))
            self.assertTrue(hasattr(image_processing, "image_std"))

    def test_image_processor_from_dict_with_kwargs(self):
        for image_processing_class in self.image_processor_list:
            image_processor = image_processing_class.from_dict(self.image_processor_dict)
            self.assertEqual(image_processor.size, {"height": 18, "width": 20})

            image_processor = image_processing_class.from_dict(self.image_processor_dict, size=42)
            self.assertEqual(image_processor.size, {"height": 42, "width": 42})

            # Previous config had dimensions in (width, height) order
            image_processor = image_processing_class.from_dict(self.image_processor_dict, size=(42, 84))
            self.assertEqual(image_processor.size, {"height": 84, "width": 42})

    def test_image_processor_preprocess_with_kwargs(self):
        for image_processing_class in self.image_processor_list:
            # Initialize image_processing
            image_processing = image_processing_class(**self.image_processor_dict)
            # create random PyTorch tensors
            image_inputs = self.image_processor_tester.prepare_image_inputs(equal_resolution=False, torchify=True)

            height = 84
            width = 42
            # Previous config had dimensions in (width, height) order
            encoded_images = image_processing(image_inputs[0], size=(width, height), return_tensors="pt").pixel_values
            self.assertEqual(
                encoded_images.shape,
                (
                    1,
                    self.image_processor_tester.num_channels,
                    height,
                    width,
                ),
            )

    @is_flaky()
    def test_call_pil(self):
        for image_processing_class in self.image_processor_list:
            # Initialize image_processing
            image_processing = image_processing_class(**self.image_processor_dict)
            # create random PIL images
            image_inputs = self.image_processor_tester.prepare_image_inputs(equal_resolution=False)
            for image in image_inputs:
                self.assertIsInstance(image, Image.Image)

            # Test not batched input
            encoded_images = image_processing(image_inputs[0], return_tensors="pt").pixel_values
            self.assertEqual(
                encoded_images.shape,
                (
                    1,
                    self.image_processor_tester.num_channels,
                    self.image_processor_tester.size["height"],
                    self.image_processor_tester.size["width"],
                ),
            )

            # Test batched
            encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
            self.assertEqual(
                encoded_images.shape,
                (
                    self.image_processor_tester.batch_size,
                    self.image_processor_tester.num_channels,
                    self.image_processor_tester.size["height"],
                    self.image_processor_tester.size["width"],
                ),
            )

    @is_flaky()
    def test_call_numpy(self):
        for image_processing_class in self.image_processor_list:
            # Initialize image_processing
            image_processing = image_processing_class(**self.image_processor_dict)
            # create random numpy tensors
            image_inputs = self.image_processor_tester.prepare_image_inputs(equal_resolution=False, numpify=True)
            for image in image_inputs:
                self.assertIsInstance(image, np.ndarray)

            # Test not batched input
            encoded_images = image_processing(image_inputs[0], return_tensors="pt").pixel_values
            self.assertEqual(
                encoded_images.shape,
                (
                    1,
                    self.image_processor_tester.num_channels,
                    self.image_processor_tester.size["height"],
                    self.image_processor_tester.size["width"],
                ),
            )

            # Test batched
            encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
            self.assertEqual(
                encoded_images.shape,
                (
                    self.image_processor_tester.batch_size,
                    self.image_processor_tester.num_channels,
                    self.image_processor_tester.size["height"],
                    self.image_processor_tester.size["width"],
                ),
            )

    @is_flaky()
    def test_call_pytorch(self):
        for image_processing_class in self.image_processor_list:
            # Initialize image_processing
            image_processing = image_processing_class(**self.image_processor_dict)
            # create random PyTorch tensors
            image_inputs = self.image_processor_tester.prepare_image_inputs(equal_resolution=False, torchify=True)
            for image in image_inputs:
                self.assertIsInstance(image, torch.Tensor)

            # Test not batched input
            encoded_images = image_processing(image_inputs[0], return_tensors="pt").pixel_values
            self.assertEqual(
                encoded_images.shape,
                (
                    1,
                    self.image_processor_tester.num_channels,
                    self.image_processor_tester.size["height"],
                    self.image_processor_tester.size["width"],
                ),
            )

            # Test batched
            encoded_images = image_processing(image_inputs, return_tensors="pt").pixel_values
            self.assertEqual(
                encoded_images.shape,
                (
                    self.image_processor_tester.batch_size,
                    self.image_processor_tester.num_channels,
                    self.image_processor_tester.size["height"],
                    self.image_processor_tester.size["width"],
                ),
            )


@require_torch
@require_vision
class DonutImageProcessingAlignAxisTest(DonutImageProcessingTest):
    def setUp(self):
        super().setUp()
        self.image_processor_tester = DonutImageProcessingTester(self, do_align_axis=True)