Added # Copied from statements

src/transformers/models/dpt/image_processing_dpt.py

@@ -283,6 +283,7 @@ def _get_pad(size, size_divisor):
         return pad(image, ((pad_size_left, pad_size_right), (pad_size_top, pad_size_bottom)), data_format=data_format)
 
     def reduce_label(self, label: ImageInput) -> np.ndarray:
+        # Copied from transformers.models.beit.image_processing_beit
         label = to_numpy_array(label)
         # Avoid using underflow conversion
         label[label == 0] = 255
@@ -429,6 +430,7 @@ def _preprocess_segmentation_map(
         return segmentation_map
 
     def __call__(self, images, segmentation_maps=None, **kwargs):
+        # Copied from transformers.models.beit.image_processing_beit
         # Overrides the `__call__` method of the `Preprocessor` class such that the images and segmentation maps can both
         # be passed in as positional arguments.
         return super().__call__(images, segmentation_maps=segmentation_maps, **kwargs)

tests/models/dpt/test_image_processing_dpt.py

@@ -91,6 +91,7 @@ def prepare_image_inputs(self, equal_resolution=False, numpify=False, torchify=F
 
 
 def prepare_semantic_single_inputs():
+    # Copied from transformers.tests.models.beit.test_image_processing_beit
     dataset = load_dataset("hf-internal-testing/fixtures_ade20k", split="test", trust_remote_code=True)
 
     image = Image.open(dataset[0]["file"])
@@ -100,6 +101,7 @@ def prepare_semantic_single_inputs():
 
 
 def prepare_semantic_batch_inputs():
+    # Copied from transformers.tests.models.beit.test_image_processing_beit
     ds = load_dataset("hf-internal-testing/fixtures_ade20k", split="test", trust_remote_code=True)
 
     image1 = Image.open(ds[0]["file"])
@@ -170,6 +172,7 @@ def test_keep_aspect_ratio(self):
         self.assertEqual(list(pixel_values.shape), [1, 3, 512, 672])
 
     def test_call_segmentation_maps(self):
+        # Copied from transformers.tests.models.beit.test_image_processing_beit
         # Initialize image_processor
         image_processor = self.image_processing_class(**self.image_processor_dict)
         # create random PyTorch tensors
@@ -276,6 +279,7 @@ def test_call_segmentation_maps(self):
         self.assertTrue(encoding["labels"].max().item() <= 255)
 
     def test_reduce_labels(self):
+        # Copied from transformers.tests.models.beit.test_image_processing_beit
         # Initialize image_processor
         image_processor = self.image_processing_class(**self.image_processor_dict)