ValueError: Expected input batch_size (182) to match target batch_size (2).

[Stevie Yeh]

When I run the deeplabv2 model on cocostuff164k, with batch_size=2,I get the error

#ValueError: Expected input batch_size (182) to match target batch_size (2).

And I check the output's of model is [2,182,41,41],and the labels'shape is [2,41,41],I think the problem is loss function.And I do not how to fix this issue.Could you help me ?Thanks

[Kazuto Nakashima]

Did you run that on multiple GPUs?

[Stevie Yeh]

Did you run that on multiple GPUs?

No,I only run that on single GPU

[Kazuto Nakashima]

Could you provide the command to reproduce the issue?

[Anton Obukhov]

It is likely you instantiate a model with 2 output classes and attempt to load coco weights.

[Kazuto Nakashima]

Thank you @toshas. I think @yejg2017 modified the code and iterated over the tensor in [2, 182, 41, 41] instead of four-scale logits 4 x [2, 182, H, W]. The error may be due to that the loss function received sliced [182, 41, 41] logit and [2, 41, 41] labels. Please check the tensor just before the loss function again. Don't switch the model to eval mode or don't remove MSC class, if you want to use the multi-scale training script intact. Please reopen this issue if someone still has the same problem and any additional information could be provided.

deeplab-pytorch/main.py

Lines 233 to 237 in 79cb390

 for logit in logits: 
     # Resize labels for {100%, 75%, 50%, Max} logits 
     _, _, H, W = logit.shape 
     labels_ = resize_labels(labels, size=(H, W)) 
     iter_loss += criterion(logit, labels_.to(device))