fix(encoder):

fix bug caused by batching in inception_mixture

gnes/encoder/video/incep_mixture.py

@@ -113,20 +113,21 @@ def encode(self, data: List['np.ndarray'], *args, **kwargs) -> np.ndarray:
                in img]
 
         @batching(concat_axis=None)
-        def _encode1(_, data):
+        def _encode1(self, data):
             _, end_points_ = self.sess.run((self.logits, self.end_points),
                                            feed_dict={self.inputs: data})
             return end_points_[self.select_layer]
 
-        v = [_ for vi in _encode1(None, img) for _ in vi]
+        v = [_ for vi in _encode1(self, img) for _ in vi]
+        print(np.array(v).shape)
 
         v_input = [v[s:e] for s, e in zip(pos_start, pos_end)]
         v_input = [(vi + [[0.0] * self.input_size] * (max_len - len(vi)))[:max_len] for vi in v_input]
         v_input = [np.array(vi, dtype=np.float32) for vi in v_input]
 
         @batching
-        def _encode2(_, data):
+        def _encode2(self, data):
             return self.sess2.run(self.mix_model.repre,
                                   feed_dict={self.mix_model.feeds: data})
 
-        return _encode2(None, v_input).astype(np.float32)
+        return _encode2(self, v_input).astype(np.float32)