[to #42322933]use Tasks.ocr_recognition

Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/10490937

modelscope/models/multi_modal/ofa/utils/constant.py

@@ -3,9 +3,9 @@ from modelscope.outputs import OutputKeys
 from modelscope.utils.constant import Tasks
 
 OFA_TASK_KEY_MAPPING = {
-    Tasks.ofa_ocr_recognition: OutputKeys.TEXT,
+    Tasks.ocr_recognition: OutputKeys.TEXT,
     Tasks.image_captioning: OutputKeys.CAPTION,
-    Tasks.summarization: OutputKeys.TEXT,
+    Tasks.text_summarization: OutputKeys.TEXT,
     Tasks.visual_question_answering: OutputKeys.TEXT,
     Tasks.visual_grounding: OutputKeys.BOXES,
     Tasks.text_classification: (OutputKeys.SCORES, OutputKeys.LABELS),

modelscope/models/multi_modal/ofa_for_all_tasks.py

@@ -27,13 +27,13 @@ __all__ = ['OfaForAllTasks']
 
 
 @MODELS.register_module(Tasks.image_captioning, module_name=Models.ofa)
-@MODELS.register_module(Tasks.ofa_ocr_recognition, module_name=Models.ofa)
+@MODELS.register_module(Tasks.ocr_recognition, module_name=Models.ofa)
 @MODELS.register_module(Tasks.visual_grounding, module_name=Models.ofa)
 @MODELS.register_module(
     Tasks.visual_question_answering, module_name=Models.ofa)
 @MODELS.register_module(Tasks.visual_entailment, module_name=Models.ofa)
 @MODELS.register_module(Tasks.image_classification, module_name=Models.ofa)
-@MODELS.register_module(Tasks.summarization, module_name=Models.ofa)
+@MODELS.register_module(Tasks.text_summarization, module_name=Models.ofa)
 @MODELS.register_module(Tasks.text_classification, module_name=Models.ofa)
 class OfaForAllTasks(TorchModel):
 
@@ -97,9 +97,9 @@ class OfaForAllTasks(TorchModel):
             'traverse': self._traverse_inference,
         }
         self.task_inference_mapping = {
-            Tasks.ofa_ocr_recognition: self._text_gen_inference,
+            Tasks.ocr_recognition: self._text_gen_inference,
             Tasks.image_captioning: self._text_gen_inference,
-            Tasks.summarization: self._text_gen_inference,
+            Tasks.text_summarization: self._text_gen_inference,
             Tasks.visual_grounding: self._visual_grounding_inference,
             Tasks.visual_entailment: inference_d[self.gen_type],
             Tasks.visual_question_answering: inference_d[self.gen_type],

modelscope/outputs.py

@@ -661,7 +661,7 @@ TASK_OUTPUTS = {
     #   "caption": "this is an image caption text."
     # }
     Tasks.image_captioning: [OutputKeys.CAPTION],
-    Tasks.ofa_ocr_recognition: [OutputKeys.TEXT],
+    Tasks.ocr_recognition: [OutputKeys.TEXT],
 
     # visual grounding result for single sample
     # {

modelscope/pipelines/multi_modal/multi_modal_embedding_pipeline.py

@@ -11,6 +11,8 @@ from modelscope.utils.logger import get_logger
 logger = get_logger()
 
 
+@PIPELINES.register_module(
+    Tasks.image_text_retrieval, module_name=Pipelines.multi_modal_embedding)
 @PIPELINES.register_module(
     Tasks.multi_modal_embedding, module_name=Pipelines.multi_modal_embedding)
 class MultiModalEmbeddingPipeline(Pipeline):

modelscope/pipelines/multi_modal/ocr_recognition_pipeline.py

@@ -16,7 +16,7 @@ logger = get_logger()
 
 
 @PIPELINES.register_module(
-    Tasks.ofa_ocr_recognition, module_name=Pipelines.ofa_ocr_recognition)
+    Tasks.ocr_recognition, module_name=Pipelines.ofa_ocr_recognition)
 class OcrRecognitionPipeline(Pipeline):
 
     def __init__(self,

modelscope/pipelines/nlp/summarization_pipeline.py

@@ -13,7 +13,7 @@ logger = get_logger()
 
 
 @PIPELINES.register_module(
-    Tasks.summarization, module_name=Pipelines.text_generation)
+    Tasks.text_summarization, module_name=Pipelines.text_generation)
 class SummarizationPipeline(Pipeline):
 
     def __init__(self,

modelscope/preprocessors/multi_modal.py

@@ -34,7 +34,7 @@ class OfaPreprocessor(Preprocessor):
         """
         super().__init__(*args, **kwargs)
         preprocess_mapping = {
-            Tasks.ofa_ocr_recognition: OfaOcrRecognitionPreprocessor,
+            Tasks.ocr_recognition: OfaOcrRecognitionPreprocessor,
             Tasks.image_captioning: OfaImageCaptioningPreprocessor,
             Tasks.visual_grounding: OfaVisualGroundingPreprocessor,
             Tasks.visual_question_answering:
@@ -42,14 +42,14 @@ class OfaPreprocessor(Preprocessor):
             Tasks.visual_entailment: OfaVisualEntailmentPreprocessor,
             Tasks.image_classification: OfaImageClassificationPreprocessor,
             Tasks.text_classification: OfaTextClassificationPreprocessor,
-            Tasks.summarization: OfaSummarizationPreprocessor,
+            Tasks.text_summarization: OfaSummarizationPreprocessor,
             Tasks.text_to_image_synthesis: OfaTextToImageSynthesisPreprocessor
         }
         input_key_mapping = {
-            Tasks.ofa_ocr_recognition: ['image'],
+            Tasks.ocr_recognition: ['image'],
             Tasks.image_captioning: ['image'],
             Tasks.image_classification: ['image'],
-            Tasks.summarization: ['text'],
+            Tasks.text_summarization: ['text'],
             Tasks.text_classification: ['text', 'text2'],
             Tasks.visual_grounding: ['image', 'text'],
             Tasks.visual_question_answering: ['image', 'text'],

modelscope/utils/constant.py

@@ -117,7 +117,7 @@ class NLPTasks(object):
     table_question_answering = 'table-question-answering'
     sentence_embedding = 'sentence-embedding'
     fill_mask = 'fill-mask'
-    summarization = 'summarization'
+    text_summarization = 'text-summarization'
     question_answering = 'question-answering'
     zero_shot_classification = 'zero-shot-classification'
     backbone = 'backbone'
@@ -151,7 +151,6 @@ class MultiModalTasks(object):
     visual_entailment = 'visual-entailment'
     video_multi_modal_embedding = 'video-multi-modal-embedding'
     image_text_retrieval = 'image-text-retrieval'
-    ofa_ocr_recognition = 'ofa-ocr-recognition'
 
 
 class TasksIODescriptions(object):

tests/pipelines/test_ofa_tasks.py

@@ -48,7 +48,7 @@ class OfaTasksTest(unittest.TestCase, DemoCompatibilityCheck):
     @unittest.skipUnless(test_level() >= 0, 'skip test in current test level')
     def test_run_with_ocr_recognize_with_name(self):
         ocr_recognize = pipeline(
-            Tasks.ofa_ocr_recognition,
+            Tasks.ocr_recognition,
             model='damo/ofa_ocr-recognition_scene_base_zh')
         result = ocr_recognize('data/test/images/image_ocr_recognition.jpg')
         print(result[OutputKeys.TEXT])
@@ -75,7 +75,7 @@ class OfaTasksTest(unittest.TestCase, DemoCompatibilityCheck):
     def test_run_with_summarization_with_model(self):
         model = Model.from_pretrained(
             'damo/ofa_summarization_gigaword_large_en')
-        ofa_pipe = pipeline(Tasks.summarization, model=model)
+        ofa_pipe = pipeline(Tasks.text_summarization, model=model)
         text = 'five-time world champion michelle kwan withdrew' + \
                'from the #### us figure skating championships on wednesday ,' + \
                ' but will petition us skating officials for the chance to ' + \
@@ -87,7 +87,7 @@ class OfaTasksTest(unittest.TestCase, DemoCompatibilityCheck):
     @unittest.skipUnless(test_level() >= 0, 'skip test in current test level')
     def test_run_with_summarization_with_name(self):
         ofa_pipe = pipeline(
-            Tasks.summarization,
+            Tasks.text_summarization,
             model='damo/ofa_summarization_gigaword_large_en')
         text = 'five-time world champion michelle kwan withdrew' + \
                'from the #### us figure skating championships on wednesday ,' + \