modify README

Signed-off-by: gexy5 <xinyu.ge@zilliz.com>

README.md

@@ -15,8 +15,8 @@ and maps vectors with labels provided by datasets used for pre-training.
 
 ## Code Example
 
-Use the pretrained Movinet model to classify and generate a vector for the given video path './archery.mp4' 
-([download](https://dl.fbaipublicfiles.com/pytorchvideo/projects/archery.mp4)).
+Use the pretrained Movinet model to classify and generate a vector for the given video path './jumpingjack.gif' 
+([download](https://github.com/tensorflow/models/raw/f8af2291cced43fc9f1d9b41ddbf772ae7b0d7d2/official/projects/movinet/files/jumpingjack.gif)).
 
  *Write the pipeline in simplified style*:
 
@@ -25,7 +25,7 @@ Use the pretrained Movinet model to classify and generate a vector for the given
 import towhee
 
 (
-    towhee.glob('./archery.mp4') 
+    towhee.glob('./jumpingjack.gif') 
           .video_decode.ffmpeg()
           .action_classification.movinet(
             model_name='movineta0', topk=5)
@@ -40,7 +40,7 @@ import towhee
 import towhee
 
 (
-    towhee.glob['path']('./archery.mp4')
+    towhee.glob['path']('./jumpingjack.gif')
           .video_decode.ffmpeg['path', 'frames']()
           .action_classification.movinet['frames', ('labels', 'scores', 'features')](
                 model_name='movineta0')
@@ -57,14 +57,14 @@ import towhee
 
 Create the operator via the following factory method
 
-***video_classification.omnivore(
+***video_classification.movinet(
 model_name='movineta0', skip_preprocess=False, classmap=None, topk=5)***
 
 **Parameters:**
 
 ​	***model_name***: *str*
 
-​	The name of pre-trained movinet model.
+​	The name of pre-trained MoViNet model.
 
 ​    Supported model names:
 - movineta0

movinet.py

@@ -71,7 +71,7 @@ class Movinet(NNOperator):
         self.transform_cfgs = get_configs(
                 side_size=172,
                 crop_size=172,
-                num_frames=30,
+                num_frames=13,
                 mean=self.input_mean,
                 std=self.input_std,
                 )
@@ -104,6 +104,7 @@ class Movinet(NNOperator):
                     )
         inputs = data.to(self.device)[None, ...]
 
+        self.model.clean_activation_buffers()
         feats = self.model.forward_features(inputs)
         features = feats.to('cpu').squeeze(0).detach().numpy()