l

Signed-off-by: xujinling <jinling.xu@zilliz.com>

bridge_former.py

@@ -8,7 +8,7 @@ import numpy
 from towhee import register
 from towhee.operator.base import NNOperator
 from towhee.types.video_frame import VideoFrame
-from towhee.models.utils.video_transforms import transform_video
+from towhee.models.utils.video_transforms import transform_video, get_configs
 from towhee.models.bridgeformer import bridge_former
 from transformers import AutoTokenizer
 
@@ -52,7 +52,15 @@ class BridgeFormer(NNOperator):
                                                 model_name=self.model_name)
         self.tokenizer = AutoTokenizer.from_pretrained('distilbert-base-uncased', TOKENIZERS_PARALLELISM=False)
 
-        self.transform_cfgs = configs(self.model_name)
+        # self.transform_cfgs = configs(self.model_name)
+        self.transform_cfgs = get_configs(
+            side_size=256,
+            crop_size=224,
+            num_frames=None,
+            mean=[0.485, 0.456, 0.406],
+            std=[0.229, 0.224, 0.225],
+        )
+        self.model.eval()
 
     def decoder_video(self,  data: List[VideoFrame]):
         # Convert list of towhee.types.Image to numpy.ndarray in float32