Fix naming issue with ImageToText pipeline (#18864)

Co-authored-by: Olivier Dehaene <olivier@huggingface.co>

docs/source/en/main_classes/pipelines.mdx

@@ -29,7 +29,7 @@ There are two categories of pipeline abstractions to be aware about:
   - [`FillMaskPipeline`]
   - [`ImageClassificationPipeline`]
   - [`ImageSegmentationPipeline`]
-  - [`Image2TextGenerationPipeline`]
+  - [`ImageToTextPipeline`]
   - [`ObjectDetectionPipeline`]
   - [`QuestionAnsweringPipeline`]
   - [`SummarizationPipeline`]
@@ -366,9 +366,9 @@ That should enable you to do all the custom code you want.
     - __call__
     - all
 
-### Image2TextGenerationPipeline
+### ImageToTextPipeline
 
-[[autodoc]] Image2TextGenerationPipeline
+[[autodoc]] ImageToTextPipeline
     - __call__
     - all
 

src/transformers/__init__.py

@@ -384,9 +384,9 @@
         "CsvPipelineDataFormat",
         "FeatureExtractionPipeline",
         "FillMaskPipeline",
-        "Image2TextGenerationPipeline",
         "ImageClassificationPipeline",
         "ImageSegmentationPipeline",
+        "ImageToTextPipeline",
         "JsonPipelineDataFormat",
         "NerPipeline",
         "ObjectDetectionPipeline",
@@ -3192,9 +3192,9 @@
         CsvPipelineDataFormat,
         FeatureExtractionPipeline,
         FillMaskPipeline,
-        Image2TextGenerationPipeline,
         ImageClassificationPipeline,
         ImageSegmentationPipeline,
+        ImageToTextPipeline,
         JsonPipelineDataFormat,
         NerPipeline,
         ObjectDetectionPipeline,

src/transformers/pipelines/__init__.py

@@ -53,9 +53,9 @@
 from .conversational import Conversation, ConversationalPipeline
 from .feature_extraction import FeatureExtractionPipeline
 from .fill_mask import FillMaskPipeline
-from .image2text_generation import Image2TextGenerationPipeline
 from .image_classification import ImageClassificationPipeline
 from .image_segmentation import ImageSegmentationPipeline
+from .image_to_text import ImageToTextPipeline
 from .object_detection import ObjectDetectionPipeline
 from .question_answering import QuestionAnsweringArgumentHandler, QuestionAnsweringPipeline
 from .table_question_answering import TableQuestionAnsweringArgumentHandler, TableQuestionAnsweringPipeline
@@ -305,8 +305,8 @@
         "default": {"model": {"pt": ("facebook/detr-resnet-50-panoptic", "fc15262")}},
         "type": "image",
     },
-    "image2text-generation": {
-        "impl": Image2TextGenerationPipeline,
+    "image-to-text": {
+        "impl": ImageToTextPipeline,
         "tf": (TFAutoModelForVision2Seq,) if is_tf_available() else (),
         "pt": (AutoModelForVision2Seq,) if is_torch_available() else (),
         "default": {

src/transformers/pipelines/image2text_generation.py → src/transformers/pipelines/image_to_text.py

@@ -26,13 +26,12 @@
 
 
 @add_end_docstrings(PIPELINE_INIT_ARGS)
-class Image2TextGenerationPipeline(Pipeline):
+class ImageToTextPipeline(Pipeline):
     """
-    Image2Text Generation pipeline using a `AutoModelForVision2Seq`. This pipeline predicts a caption for a given
-    image.
+    Image To Text pipeline using a `AutoModelForVision2Seq`. This pipeline predicts a caption for a given image.
 
-    This image to text generation pipeline can currently be loaded from pipeline() using the following task identifier:
-    "image2text-generation".
+    This image to text pipeline can currently be loaded from pipeline() using the following task identifier:
+    "image-to-text".
 
     See the list of available models on
     [huggingface.co/models](https://huggingface.co/models?pipeline_tag=image-to-text).

tests/pipelines/test_pipelines_image2text_generation.py → tests/pipelines/test_pipelines_image_to_text.py

@@ -33,12 +33,12 @@ def open(*args, **kwargs):
 
 @is_pipeline_test
 @require_vision
-class Image2TextGenerationPipelineTests(unittest.TestCase, metaclass=PipelineTestCaseMeta):
+class ImageToTextPipelineTests(unittest.TestCase, metaclass=PipelineTestCaseMeta):
     model_mapping = MODEL_FOR_VISION_2_SEQ_MAPPING
     tf_model_mapping = TF_MODEL_FOR_VISION_2_SEQ_MAPPING
 
     def get_test_pipeline(self, model, tokenizer, feature_extractor):
-        pipe = pipeline("image2text-generation", model=model, tokenizer=tokenizer, feature_extractor=feature_extractor)
+        pipe = pipeline("image-to-text", model=model, tokenizer=tokenizer, feature_extractor=feature_extractor)
         examples = [
             Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png"),
             "./tests/fixtures/tests_samples/COCO/000000039769.png",
@@ -57,7 +57,7 @@ def run_pipeline_test(self, pipe, examples):
 
     @require_tf
     def test_small_model_tf(self):
-        pipe = pipeline("image2text-generation", model="hf-internal-testing/tiny-random-vit-gpt2")
+        pipe = pipeline("image-to-text", model="hf-internal-testing/tiny-random-vit-gpt2")
         image = "./tests/fixtures/tests_samples/COCO/000000039769.png"
 
         outputs = pipe(image)
@@ -104,7 +104,7 @@ def test_small_model_tf(self):
 
     @require_torch
     def test_small_model_pt(self):
-        pipe = pipeline("image2text-generation", model="hf-internal-testing/tiny-random-vit-gpt2")
+        pipe = pipeline("image-to-text", model="hf-internal-testing/tiny-random-vit-gpt2")
         image = "./tests/fixtures/tests_samples/COCO/000000039769.png"
 
         outputs = pipe(image)
@@ -137,7 +137,7 @@ def test_small_model_pt(self):
     @slow
     @require_torch
     def test_large_model_pt(self):
-        pipe = pipeline("image2text-generation", model="ydshieh/vit-gpt2-coco-en")
+        pipe = pipeline("image-to-text", model="ydshieh/vit-gpt2-coco-en")
         image = "./tests/fixtures/tests_samples/COCO/000000039769.png"
 
         outputs = pipe(image)
@@ -155,7 +155,7 @@ def test_large_model_pt(self):
     @slow
     @require_tf
     def test_large_model_tf(self):
-        pipe = pipeline("image2text-generation", model="ydshieh/vit-gpt2-coco-en")
+        pipe = pipeline("image-to-text", model="ydshieh/vit-gpt2-coco-en")
         image = "./tests/fixtures/tests_samples/COCO/000000039769.png"
 
         outputs = pipe(image)