Fix SegformerForImageClassification (huggingface#15895)

NielsRogge · Niels Rogge · web-flow · commit 89be34c36c1a · 2022-03-02T21:57:39.000+01:00
* Fix reshape

* Apply suggestion from code review

Co-authored-by: Niels Rogge &lt;nielsrogge@Nielss-MacBook-Pro.local&gt;
diff --git a/src/transformers/models/segformer/modeling_segformer.py b/src/transformers/models/segformer/modeling_segformer.py
@@ -579,8 +579,11 @@ def forward(
 
         sequence_output = outputs[0]
 
-        # reshape last hidden states to (batch_size, height*width, hidden_size)
+        # convert last hidden states to (batch_size, height*width, hidden_size)
         batch_size = sequence_output.shape[0]
+        if self.config.reshape_last_stage:
+            # (batch_size, num_channels, height, width) -> (batch_size, height, width, num_channels)
+            sequence_output = sequence_output.permute(0, 2, 3, 1)
         sequence_output = sequence_output.reshape(batch_size, -1, self.config.hidden_sizes[-1])
 
         # global average pooling
@@ -660,10 +663,19 @@ def __init__(self, config):
         self.dropout = nn.Dropout(config.classifier_dropout_prob)
         self.classifier = nn.Conv2d(config.decoder_hidden_size, config.num_labels, kernel_size=1)
 
+        self.config = config
+
     def forward(self, encoder_hidden_states):
-        batch_size, _, _, _ = encoder_hidden_states[-1].shape
+        batch_size = encoder_hidden_states[-1].shape[0]
+
         all_hidden_states = ()
         for encoder_hidden_state, mlp in zip(encoder_hidden_states, self.linear_c):
+            if self.config.reshape_last_stage is False and encoder_hidden_state.ndim == 3:
+                height = width = int(math.sqrt(encoder_hidden_state.shape[-1]))
+                encoder_hidden_state = (
+                    encoder_hidden_state.reshape(batch_size, height, width, -1).permute(0, 3, 1, 2).contiguous()
+                )
+
             # unify channel dimension
             height, width = encoder_hidden_state.shape[2], encoder_hidden_state.shape[3]
             encoder_hidden_state = mlp(encoder_hidden_state)
