[to #44657982] fix some demo problems

Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/10083156

modelscope/models/multi_modal/ofa_for_all_tasks.py

@@ -152,8 +152,8 @@ class OfaForAllTasks(TorchModel):
         region_tensor[:, ::2] /= input['w_resize_ratios']
         region_tensor[:, 1::2] /= input['h_resize_ratios']
         return {
-            OutputKeys.BOXES: move_to_device(region_tensor,
-                                             torch.device('cpu')),
+            OutputKeys.BOXES:
+            move_to_device(region_tensor, torch.device('cpu')).tolist(),
             OutputKeys.SCORES: [1.0] * region_tensor.shape[0]
         }
 

modelscope/pipelines/audio/linear_aec_pipeline.py

@@ -51,7 +51,7 @@ class LinearAECPipeline(Pipeline):
 
     When invoke the class with pipeline.__call__(), you should provide two params:
         Dict[str, Any]
-            the path of wav files，eg:{
+            the path of wav files, eg:{
             "nearend_mic": "/your/data/near_end_mic_audio.wav",
             "farend_speech": "/your/data/far_end_speech_audio.wav"}
         output_path (str, optional): "/your/output/audio_after_aec.wav"

modelscope/preprocessors/multi_modal.py

@@ -8,6 +8,7 @@ from PIL import Image
 from modelscope.hub.snapshot_download import snapshot_download
 from modelscope.metainfo import Preprocessors
 from modelscope.pipelines.base import Input
+from modelscope.preprocessors import load_image
 from modelscope.utils.config import Config
 from modelscope.utils.constant import Fields, ModeKeys, ModelFile, Tasks
 from .base import Preprocessor
@@ -137,7 +138,7 @@ class MPlugPreprocessor(Preprocessor):
     def image_open(self, path: str) -> Tuple[Image.Image, int]:
         if path not in self._image_map:
             index = len(self._image_map)
-            self._image_map[path] = (Image.open(path), index)
+            self._image_map[path] = (load_image(path), index)
         return self._image_map[path]
 
     def __call__(

modelscope/utils/demo_utils.py

@@ -236,7 +236,7 @@ def postprocess(req, resp):
             _, img_encode = cv2.imencode('.' + file_type, content)
             img_bytes = img_encode.tobytes()
             return type(img_bytes)
-        elif file_type == 'wav':
+        else:
             out_mem_file = io.BytesIO()
             out_mem_file.write(new_resp.get(output_key))
             return type(out_mem_file)

tests/pipelines/test_automatic_speech_recognition.py

@@ -22,6 +22,9 @@ URL_FILE = 'https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audi
 LITTLE_TESTSETS_FILE = 'data_aishell.tar.gz'
 LITTLE_TESTSETS_URL = 'https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/datasets/data_aishell.tar.gz'
 
+TFRECORD_TESTSETS_FILE = 'tfrecord.tar.gz'
+TFRECORD_TESTSETS_URL = 'https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/datasets/tfrecord.tar.gz'
+
 
 class AutomaticSpeechRecognitionTest(unittest.TestCase,
                                      DemoCompatibilityCheck):