Update old existing feature extractor references (#24552)

* Update old existing feature extractor references

* Typo

* Apply suggestions from code review

* Apply suggestions from code review

* Apply suggestions from code review

* Address comments from review - update 'feature extractor'
Co-authored by: Yih-Dar <2521628+ydshieh@users.noreply.github.com>

tests/models/swin/test_modeling_tf_swin.py

@@ -45,7 +45,7 @@ if is_tf_available():
 if is_vision_available():
     from PIL import Image
 
-    from transformers import AutoFeatureExtractor
+    from transformers import AutoImageProcessor
 
 
 class TFSwinModelTester:
@@ -382,9 +382,9 @@ class TFSwinModelTest(TFModelTesterMixin, PipelineTesterMixin, unittest.TestCase
 @require_tf
 class TFSwinModelIntegrationTest(unittest.TestCase):
     @cached_property
-    def default_feature_extractor(self):
+    def default_image_processor(self):
         return (
-            AutoFeatureExtractor.from_pretrained("microsoft/swin-tiny-patch4-window7-224")
+            AutoImageProcessor.from_pretrained("microsoft/swin-tiny-patch4-window7-224")
             if is_vision_available()
             else None
         )
@@ -392,10 +392,10 @@ class TFSwinModelIntegrationTest(unittest.TestCase):
     @slow
     def test_inference_image_classification_head(self):
         model = TFSwinForImageClassification.from_pretrained("microsoft/swin-tiny-patch4-window7-224")
-        feature_extractor = self.default_feature_extractor
+        image_processor = self.default_image_processor
 
         image = Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png")
-        inputs = feature_extractor(images=image, return_tensors="tf")
+        inputs = image_processor(images=image, return_tensors="tf")
 
         # forward pass
         outputs = model(inputs)

tests/models/swinv2/test_modeling_swinv2.py

@@ -36,7 +36,7 @@ if is_torch_available():
 if is_vision_available():
     from PIL import Image
 
-    from transformers import AutoFeatureExtractor
+    from transformers import AutoImageProcessor
 
 
 class Swinv2ModelTester:
@@ -412,9 +412,9 @@ class Swinv2ModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
 @require_torch
 class Swinv2ModelIntegrationTest(unittest.TestCase):
     @cached_property
-    def default_feature_extractor(self):
+    def default_image_processor(self):
         return (
-            AutoFeatureExtractor.from_pretrained("microsoft/swinv2-tiny-patch4-window8-256")
+            AutoImageProcessor.from_pretrained("microsoft/swinv2-tiny-patch4-window8-256")
             if is_vision_available()
             else None
         )
@@ -424,10 +424,10 @@ class Swinv2ModelIntegrationTest(unittest.TestCase):
         model = Swinv2ForImageClassification.from_pretrained("microsoft/swinv2-tiny-patch4-window8-256").to(
             torch_device
         )
-        feature_extractor = self.default_feature_extractor
+        image_processor = self.default_image_processor
 
         image = Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png")
-        inputs = feature_extractor(images=image, return_tensors="pt").to(torch_device)
+        inputs = image_processor(images=image, return_tensors="pt").to(torch_device)
 
         # forward pass
         with torch.no_grad():

tests/models/table_transformer/test_modeling_table_transformer.py

@@ -39,7 +39,7 @@ if is_timm_available():
 if is_vision_available():
     from PIL import Image
 
-    from transformers import AutoFeatureExtractor
+    from transformers import AutoImageProcessor
 
 
 class TableTransformerModelTester:
@@ -501,13 +501,13 @@ def prepare_img():
 @slow
 class TableTransformerModelIntegrationTests(unittest.TestCase):
     def test_table_detection(self):
-        feature_extractor = AutoFeatureExtractor.from_pretrained("microsoft/table-transformer-detection")
+        image_processor = AutoImageProcessor.from_pretrained("microsoft/table-transformer-detection")
         model = TableTransformerForObjectDetection.from_pretrained("microsoft/table-transformer-detection")
         model.to(torch_device)
 
         file_path = hf_hub_download(repo_id="nielsr/example-pdf", repo_type="dataset", filename="example_pdf.png")
         image = Image.open(file_path).convert("RGB")
-        inputs = feature_extractor(image, return_tensors="pt").to(torch_device)
+        inputs = image_processor(image, return_tensors="pt").to(torch_device)
 
         # forward pass
         with torch.no_grad():

tests/models/timesformer/test_modeling_timesformer.py

@@ -45,7 +45,7 @@ if is_torch_available():
 
 
 if is_vision_available():
-    from transformers import VideoMAEFeatureExtractor
+    from transformers import VideoMAEImageProcessor
 
 
 class TimesformerModelTester:
@@ -339,10 +339,10 @@ def prepare_video():
 @require_vision
 class TimesformerModelIntegrationTest(unittest.TestCase):
     @cached_property
-    def default_feature_extractor(self):
+    def default_image_processor(self):
         # logits were tested with a different mean and std, so we use the same here
         return (
-            VideoMAEFeatureExtractor(image_mean=[0.5, 0.5, 0.5], image_std=[0.5, 0.5, 0.5])
+            VideoMAEImageProcessor(image_mean=[0.5, 0.5, 0.5], image_std=[0.5, 0.5, 0.5])
             if is_vision_available()
             else None
         )
@@ -353,9 +353,9 @@ class TimesformerModelIntegrationTest(unittest.TestCase):
             torch_device
         )
 
-        feature_extractor = self.default_feature_extractor
+        image_processor = self.default_image_processor
         video = prepare_video()
-        inputs = feature_extractor(video[:8], return_tensors="pt").to(torch_device)
+        inputs = image_processor(video[:8], return_tensors="pt").to(torch_device)
 
         # forward pass
         with torch.no_grad():

tests/models/tvlt/test_modeling_tvlt.py

@@ -564,7 +564,7 @@ def prepare_audio(num_samples=1):
 @require_vision
 class TvltModelIntegrationTest(unittest.TestCase):
     @cached_property
-    def default_feature_extractor(self):
+    def default_processors(self):
         # logits were tested with a different mean and std, so we use the same here
         return (
             TvltImageProcessor() if is_vision_available() else None,
@@ -574,7 +574,7 @@ class TvltModelIntegrationTest(unittest.TestCase):
     def test_inference_for_base_model(self):
         model = TvltModel.from_pretrained("ZinengTang/tvlt-base").to(torch_device)
 
-        image_processor, audio_feature_extractor = self.default_feature_extractor
+        image_processor, audio_feature_extractor = self.default_processors
         video = prepare_video()
         audio = prepare_audio()
         video_inputs = image_processor(video, return_tensors="pt").to(torch_device)
@@ -596,7 +596,7 @@ class TvltModelIntegrationTest(unittest.TestCase):
     def test_inference_for_pretraining(self):
         model = TvltForPreTraining.from_pretrained("ZinengTang/tvlt-base").to(torch_device)
 
-        image_processor, audio_feature_extractor = self.default_feature_extractor
+        image_processor, audio_feature_extractor = self.default_processors
         video = prepare_video()
         video_mixed = prepare_video()
         audio = prepare_audio()

tests/models/van/test_modeling_van.py

@@ -42,7 +42,7 @@ if is_torch_available():
 if is_vision_available():
     from PIL import Image
 
-    from transformers import AutoFeatureExtractor
+    from transformers import AutoImageProcessor
 
 
 class VanModelTester:
@@ -254,16 +254,16 @@ def prepare_img():
 @require_vision
 class VanModelIntegrationTest(unittest.TestCase):
     @cached_property
-    def default_feature_extractor(self):
-        return AutoFeatureExtractor.from_pretrained(VAN_PRETRAINED_MODEL_ARCHIVE_LIST[0])
+    def default_image_processor(self):
+        return AutoImageProcessor.from_pretrained(VAN_PRETRAINED_MODEL_ARCHIVE_LIST[0])
 
     @slow
     def test_inference_image_classification_head(self):
         model = VanForImageClassification.from_pretrained(VAN_PRETRAINED_MODEL_ARCHIVE_LIST[0]).to(torch_device)
 
-        feature_extractor = self.default_feature_extractor
+        image_processor = self.default_image_processor
         image = prepare_img()
-        inputs = feature_extractor(images=image, return_tensors="pt").to(torch_device)
+        inputs = image_processor(images=image, return_tensors="pt").to(torch_device)
 
         # forward pass
         with torch.no_grad():

tests/models/videomae/test_modeling_videomae.py

@@ -46,7 +46,7 @@ if is_torch_available():
 
 
 if is_vision_available():
-    from transformers import VideoMAEFeatureExtractor
+    from transformers import VideoMAEImageProcessor
 
 
 class VideoMAEModelTester:
@@ -359,10 +359,10 @@ def prepare_video():
 @require_vision
 class VideoMAEModelIntegrationTest(unittest.TestCase):
     @cached_property
-    def default_feature_extractor(self):
+    def default_image_processor(self):
         # logits were tested with a different mean and std, so we use the same here
         return (
-            VideoMAEFeatureExtractor(image_mean=[0.5, 0.5, 0.5], image_std=[0.5, 0.5, 0.5])
+            VideoMAEImageProcessor(image_mean=[0.5, 0.5, 0.5], image_std=[0.5, 0.5, 0.5])
             if is_vision_available()
             else None
         )
@@ -373,9 +373,9 @@ class VideoMAEModelIntegrationTest(unittest.TestCase):
             torch_device
         )
 
-        feature_extractor = self.default_feature_extractor
+        image_processor = self.default_image_processor
         video = prepare_video()
-        inputs = feature_extractor(video, return_tensors="pt").to(torch_device)
+        inputs = image_processor(video, return_tensors="pt").to(torch_device)
 
         # forward pass
         with torch.no_grad():
@@ -393,9 +393,9 @@ class VideoMAEModelIntegrationTest(unittest.TestCase):
     def test_inference_for_pretraining(self):
         model = VideoMAEForPreTraining.from_pretrained("MCG-NJU/videomae-base-short").to(torch_device)
 
-        feature_extractor = self.default_feature_extractor
+        image_processor = self.default_image_processor
         video = prepare_video()
-        inputs = feature_extractor(video, return_tensors="pt").to(torch_device)
+        inputs = image_processor(video, return_tensors="pt").to(torch_device)
 
         # add boolean mask, indicating which patches to mask
         local_path = hf_hub_download(repo_id="hf-internal-testing/bool-masked-pos", filename="bool_masked_pos.pt")

tests/models/vision_encoder_decoder/test_modeling_flax_vision_encoder_decoder.py

@@ -48,7 +48,7 @@ if is_torch_available():
 if is_vision_available():
     from PIL import Image
 
-    from transformers import ViTFeatureExtractor
+    from transformers import ViTImageProcessor
 
 
 @require_flax
@@ -462,12 +462,12 @@ class FlaxViT2GPT2ModelIntegrationTest(unittest.TestCase):
     def test_inference_coco_en(self):
         loc = "ydshieh/vit-gpt2-coco-en"
 
-        feature_extractor = ViTFeatureExtractor.from_pretrained(loc)
+        image_processor = ViTImageProcessor.from_pretrained(loc)
         tokenizer = AutoTokenizer.from_pretrained(loc)
         model = FlaxVisionEncoderDecoderModel.from_pretrained(loc)
 
         img = prepare_img()
-        pixel_values = feature_extractor(images=img, return_tensors="np").pixel_values
+        pixel_values = image_processor(images=img, return_tensors="np").pixel_values
 
         decoder_input_ids = np.array([[model.config.decoder_start_token_id]])
         logits = model(pixel_values, decoder_input_ids)[0]

tests/models/vision_encoder_decoder/test_modeling_tf_vision_encoder_decoder.py

@@ -45,7 +45,7 @@ if is_tf_available():
 
     from transformers import (
         AutoConfig,
-        AutoFeatureExtractor,
+        AutoImageProcessor,
         AutoTokenizer,
         TFAutoModel,
         TFAutoModelForCausalLM,
@@ -64,7 +64,7 @@ if is_torch_available():
 if is_vision_available():
     from PIL import Image
 
-    from transformers import ViTFeatureExtractor
+    from transformers import ViTImageProcessor
 
 
 @require_tf
@@ -828,11 +828,11 @@ class TFVisionEncoderDecoderModelSaveLoadTests(unittest.TestCase):
         load_weight_prefix = TFVisionEncoderDecoderModel.load_weight_prefix
 
         config = self.get_encoder_decoder_config()
-        feature_extractor = AutoFeatureExtractor.from_pretrained("google/vit-base-patch16-224-in21k")
+        image_processor = AutoImageProcessor.from_pretrained("google/vit-base-patch16-224-in21k")
         decoder_tokenizer = AutoTokenizer.from_pretrained("gpt2")
 
         img = prepare_img()
-        pixel_values = feature_extractor(images=img, return_tensors="tf").pixel_values
+        pixel_values = image_processor(images=img, return_tensors="tf").pixel_values
         decoder_input_ids = decoder_tokenizer("Linda Davis", return_tensors="tf").input_ids
 
         with tempfile.TemporaryDirectory() as tmp_dirname:
@@ -893,13 +893,13 @@ class TFViT2GPT2ModelIntegrationTest(unittest.TestCase):
     def test_inference_coco_en(self):
         loc = "ydshieh/vit-gpt2-coco-en"
 
-        feature_extractor = ViTFeatureExtractor.from_pretrained(loc)
+        image_processor = ViTImageProcessor.from_pretrained(loc)
         tokenizer = AutoTokenizer.from_pretrained(loc)
         model = TFVisionEncoderDecoderModel.from_pretrained(loc)
 
         # We will verify our results on an image of cute cats
         img = Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png")
-        pixel_values = feature_extractor(images=img, return_tensors="tf").pixel_values
+        pixel_values = image_processor(images=img, return_tensors="tf").pixel_values
 
         decoder_input_ids = tf.constant([[model.config.decoder_start_token_id]])
 

tests/models/vision_encoder_decoder/test_modeling_vision_encoder_decoder.py

@@ -62,7 +62,7 @@ if is_vision_available():
     import PIL
     from PIL import Image
 
-    from transformers import ViTFeatureExtractor
+    from transformers import ViTImageProcessor
 
 
 @require_torch
@@ -749,7 +749,7 @@ class ViT2GPT2ModelIntegrationTest(unittest.TestCase):
     def test_inference_coco_en(self):
         loc = "ydshieh/vit-gpt2-coco-en"
 
-        feature_extractor = ViTFeatureExtractor.from_pretrained(loc)
+        image_processor = ViTImageProcessor.from_pretrained(loc)
         tokenizer = AutoTokenizer.from_pretrained(loc)
         model = VisionEncoderDecoderModel.from_pretrained(loc)
         model.to(torch_device)
@@ -757,7 +757,7 @@ class ViT2GPT2ModelIntegrationTest(unittest.TestCase):
 
         # We will verify our results on an image of cute cats
         img = Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png")
-        pixel_values = feature_extractor(images=img, return_tensors="pt").pixel_values.to(torch_device)
+        pixel_values = image_processor(images=img, return_tensors="pt").pixel_values.to(torch_device)
 
         decoder_input_ids = torch.tensor([[model.config.decoder_start_token_id]]).to(torch_device)
 

tests/models/vision_text_dual_encoder/test_processor_vision_text_dual_encoder.py

@@ -170,10 +170,10 @@ class VisionTextDualEncoderProcessorTest(unittest.TestCase):
         self.assertListEqual(decoded_tok, decoded_processor)
 
     def test_model_input_names(self):
-        feature_extractor = self.get_image_processor()
+        image_processor = self.get_image_processor()
         tokenizer = self.get_tokenizer()
 
-        processor = VisionTextDualEncoderProcessor(tokenizer=tokenizer, feature_extractor=feature_extractor)
+        processor = VisionTextDualEncoderProcessor(tokenizer=tokenizer, image_processor=image_processor)
 
         input_str = "lower newer"
         image_input = self.prepare_image_inputs()

tests/models/vit/test_modeling_tf_vit.py

@@ -38,7 +38,7 @@ if is_tf_available():
 if is_vision_available():
     from PIL import Image
 
-    from transformers import ViTFeatureExtractor
+    from transformers import ViTImageProcessor
 
 
 class TFViTModelTester:
@@ -228,16 +228,16 @@ def prepare_img():
 @require_vision
 class TFViTModelIntegrationTest(unittest.TestCase):
     @cached_property
-    def default_feature_extractor(self):
-        return ViTFeatureExtractor.from_pretrained("google/vit-base-patch16-224") if is_vision_available() else None
+    def default_image_processor(self):
+        return ViTImageProcessor.from_pretrained("google/vit-base-patch16-224") if is_vision_available() else None
 
     @slow
     def test_inference_image_classification_head(self):
         model = TFViTForImageClassification.from_pretrained("google/vit-base-patch16-224")
 
-        feature_extractor = self.default_feature_extractor
+        image_processor = self.default_image_processor
         image = prepare_img()
-        inputs = feature_extractor(images=image, return_tensors="tf")
+        inputs = image_processor(images=image, return_tensors="tf")
 
         # forward pass
         outputs = model(**inputs)

tests/models/vit/test_modeling_vit.py

@@ -45,7 +45,7 @@ if is_torch_available():
 if is_vision_available():
     from PIL import Image
 
-    from transformers import ViTFeatureExtractor
+    from transformers import ViTImageProcessor
 
 
 class ViTModelTester:
@@ -264,16 +264,16 @@ def prepare_img():
 @require_vision
 class ViTModelIntegrationTest(unittest.TestCase):
     @cached_property
-    def default_feature_extractor(self):
-        return ViTFeatureExtractor.from_pretrained("google/vit-base-patch16-224") if is_vision_available() else None
+    def default_image_processor(self):
+        return ViTImageProcessor.from_pretrained("google/vit-base-patch16-224") if is_vision_available() else None
 
     @slow
     def test_inference_image_classification_head(self):
         model = ViTForImageClassification.from_pretrained("google/vit-base-patch16-224").to(torch_device)
 
-        feature_extractor = self.default_feature_extractor
+        image_processor = self.default_image_processor
         image = prepare_img()
-        inputs = feature_extractor(images=image, return_tensors="pt").to(torch_device)
+        inputs = image_processor(images=image, return_tensors="pt").to(torch_device)
 
         # forward pass
         with torch.no_grad():
@@ -295,9 +295,9 @@ class ViTModelIntegrationTest(unittest.TestCase):
         # to visualize self-attention on higher resolution images.
         model = ViTModel.from_pretrained("facebook/dino-vits8").to(torch_device)
 
-        feature_extractor = ViTFeatureExtractor.from_pretrained("facebook/dino-vits8", size=480)
+        image_processor = ViTImageProcessor.from_pretrained("facebook/dino-vits8", size=480)
         image = prepare_img()
-        inputs = feature_extractor(images=image, return_tensors="pt")
+        inputs = image_processor(images=image, return_tensors="pt")
         pixel_values = inputs.pixel_values.to(torch_device)
 
         # forward pass
@@ -322,10 +322,10 @@ class ViTModelIntegrationTest(unittest.TestCase):
         A small test to make sure that inference work in half precision without any problem.
         """
         model = ViTModel.from_pretrained("facebook/dino-vits8", torch_dtype=torch.float16, device_map="auto")
-        feature_extractor = self.default_feature_extractor
+        image_processor = self.default_image_processor
 
         image = prepare_img()
-        inputs = feature_extractor(images=image, return_tensors="pt")
+        inputs = image_processor(images=image, return_tensors="pt")
         pixel_values = inputs.pixel_values.to(torch_device)
 
         # forward pass to make sure inference works in fp16

tests/models/vit_hybrid/test_modeling_vit_hybrid.py

@@ -243,7 +243,7 @@ def prepare_img():
 @require_vision
 class ViTModelIntegrationTest(unittest.TestCase):
     @cached_property
-    def default_feature_extractor(self):
+    def default_image_processor(self):
         return (
             ViTHybridImageProcessor.from_pretrained(VIT_HYBRID_PRETRAINED_MODEL_ARCHIVE_LIST[0])
             if is_vision_available()
@@ -256,9 +256,9 @@ class ViTModelIntegrationTest(unittest.TestCase):
             torch_device
         )
 
-        feature_extractor = self.default_feature_extractor
+        image_processor = self.default_image_processor
         image = prepare_img()
-        inputs = feature_extractor(images=image, return_tensors="pt").to(torch_device)
+        inputs = image_processor(images=image, return_tensors="pt").to(torch_device)
 
         # forward pass
         with torch.no_grad():
@@ -275,12 +275,12 @@ class ViTModelIntegrationTest(unittest.TestCase):
     @slow
     @require_accelerate
     def test_accelerate_inference(self):
-        feature_extractor = ViTHybridImageProcessor.from_pretrained("google/vit-hybrid-base-bit-384")
+        image_processor = ViTHybridImageProcessor.from_pretrained("google/vit-hybrid-base-bit-384")
         model = ViTHybridForImageClassification.from_pretrained("google/vit-hybrid-base-bit-384", device_map="auto")
 
         image = prepare_img()
 
-        inputs = feature_extractor(images=image, return_tensors="pt")
+        inputs = image_processor(images=image, return_tensors="pt")
         outputs = model(**inputs)
         logits = outputs.logits
         # model predicts one of the 1000 ImageNet classes

tests/models/vit_mae/test_modeling_tf_vit_mae.py

@@ -46,7 +46,7 @@ if is_tf_available():
 if is_vision_available():
     from PIL import Image
 
-    from transformers import ViTFeatureExtractor
+    from transformers import ViTImageProcessor
 
 
 class TFViTMAEModelTester:
@@ -424,8 +424,8 @@ def prepare_img():
 @require_vision
 class TFViTMAEModelIntegrationTest(unittest.TestCase):
     @cached_property
-    def default_feature_extractor(self):
-        return ViTFeatureExtractor.from_pretrained("facebook/vit-mae-base") if is_vision_available() else None
+    def default_image_processor(self):
+        return ViTImageProcessor.from_pretrained("facebook/vit-mae-base") if is_vision_available() else None
 
     @slow
     def test_inference_for_pretraining(self):
@@ -434,9 +434,9 @@ class TFViTMAEModelIntegrationTest(unittest.TestCase):
 
         model = TFViTMAEForPreTraining.from_pretrained("facebook/vit-mae-base")
 
-        feature_extractor = self.default_feature_extractor
+        image_processor = self.default_image_processor
         image = prepare_img()
-        inputs = feature_extractor(images=image, return_tensors="tf")
+        inputs = image_processor(images=image, return_tensors="tf")
 
         # prepare a noise vector that will be also used for testing the TF model
         # (this way we can ensure that the PT and TF models operate on the same inputs)

tests/models/vit_mae/test_modeling_vit_mae.py

@@ -42,7 +42,7 @@ if is_torch_available():
 if is_vision_available():
     from PIL import Image
 
-    from transformers import ViTFeatureExtractor
+    from transformers import ViTImageProcessor
 
 
 class ViTMAEModelTester:
@@ -296,8 +296,8 @@ def prepare_img():
 @require_vision
 class ViTMAEModelIntegrationTest(unittest.TestCase):
     @cached_property
-    def default_feature_extractor(self):
-        return ViTFeatureExtractor.from_pretrained("facebook/vit-mae-base") if is_vision_available() else None
+    def default_image_processor(self):
+        return ViTImageProcessor.from_pretrained("facebook/vit-mae-base") if is_vision_available() else None
 
     @slow
     def test_inference_for_pretraining(self):
@@ -306,9 +306,9 @@ class ViTMAEModelIntegrationTest(unittest.TestCase):
 
         model = ViTMAEForPreTraining.from_pretrained("facebook/vit-mae-base").to(torch_device)
 
-        feature_extractor = self.default_feature_extractor
+        image_processor = self.default_image_processor
         image = prepare_img()
-        inputs = feature_extractor(images=image, return_tensors="pt").to(torch_device)
+        inputs = image_processor(images=image, return_tensors="pt").to(torch_device)
 
         # prepare a noise vector that will be also used for testing the TF model
         # (this way we can ensure that the PT and TF models operate on the same inputs)

tests/models/vit_msn/test_modeling_vit_msn.py

@@ -38,7 +38,7 @@ if is_torch_available():
 if is_vision_available():
     from PIL import Image
 
-    from transformers import ViTFeatureExtractor
+    from transformers import ViTImageProcessor
 
 
 class ViTMSNModelTester:
@@ -220,17 +220,17 @@ def prepare_img():
 @require_vision
 class ViTMSNModelIntegrationTest(unittest.TestCase):
     @cached_property
-    def default_feature_extractor(self):
-        return ViTFeatureExtractor.from_pretrained("facebook/vit-msn-small") if is_vision_available() else None
+    def default_image_processor(self):
+        return ViTImageProcessor.from_pretrained("facebook/vit-msn-small") if is_vision_available() else None
 
     @slow
     def test_inference_image_classification_head(self):
         torch.manual_seed(2)
         model = ViTMSNForImageClassification.from_pretrained("facebook/vit-msn-small").to(torch_device)
 
-        feature_extractor = self.default_feature_extractor
+        image_processor = self.default_image_processor
         image = prepare_img()
-        inputs = feature_extractor(images=image, return_tensors="pt").to(torch_device)
+        inputs = image_processor(images=image, return_tensors="pt").to(torch_device)
 
         # forward pass
         with torch.no_grad():

tests/models/yolos/test_modeling_yolos.py

@@ -38,7 +38,7 @@ if is_torch_available():
 if is_vision_available():
     from PIL import Image
 
-    from transformers import AutoFeatureExtractor
+    from transformers import AutoImageProcessor
 
 
 class YolosModelTester:
@@ -345,16 +345,16 @@ def prepare_img():
 @require_vision
 class YolosModelIntegrationTest(unittest.TestCase):
     @cached_property
-    def default_feature_extractor(self):
-        return AutoFeatureExtractor.from_pretrained("hustvl/yolos-small") if is_vision_available() else None
+    def default_image_processor(self):
+        return AutoImageProcessor.from_pretrained("hustvl/yolos-small") if is_vision_available() else None
 
     @slow
     def test_inference_object_detection_head(self):
         model = YolosForObjectDetection.from_pretrained("hustvl/yolos-small").to(torch_device)
 
-        feature_extractor = self.default_feature_extractor
+        image_processor = self.default_image_processor
         image = prepare_img()
-        inputs = feature_extractor(images=image, return_tensors="pt").to(torch_device)
+        inputs = image_processor(images=image, return_tensors="pt").to(torch_device)
 
         # forward pass
         with torch.no_grad():
@@ -375,7 +375,7 @@ class YolosModelIntegrationTest(unittest.TestCase):
         self.assertTrue(torch.allclose(outputs.pred_boxes[0, :3, :3], expected_slice_boxes, atol=1e-4))
 
         # verify postprocessing
-        results = feature_extractor.post_process_object_detection(
+        results = image_processor.post_process_object_detection(
             outputs, threshold=0.3, target_sizes=[image.size[::-1]]
         )[0]
         expected_scores = torch.tensor([0.9994, 0.9790, 0.9964, 0.9972, 0.9861]).to(torch_device)