Error while loading DeepLabV3Plus-ResNet50 model from checkpoint with --separable_conv flag about deeplabv3plus-pytorch HOT 4 CLOSED

Hi @free-bit , thanks for your suggestions. The readme was updated.

from deeplabv3plus-pytorch.

@VainF I encountered the opposite situation.I get weight by -separable_conv,Then I encountered problems during the test.

RuntimeError: Error(s) in loading state_dict for DeepLabV3:
Missing key(s) in state_dict: "classifier.aspp.convs.1.0.weight", "classifier.aspp.convs.2.0.weight", "classifier.aspp.convs.3.0.weight", "classifier.classifier.0.weight".
Unexpected key(s) in state_dict: "classifier.aspp.convs.1.0.body.0.weight", "classifier.aspp.convs.1.0.body.1.weight", "classifier.aspp.convs.2.0.body.0.weight", "classifier.aspp.convs.2.0.body.1.weight", "classifier.aspp.convs.3.0.body.0.weight", "classifier.aspp.convs.3.0.body.1.weight", "classifier.classifier.0.body.0.weight", "classifier.classifier.0.body.1.weight".

I think it's because the weight name was changed during training，How to modify the code to test after -separable_conv?

from deeplabv3plus-pytorch.

Hi @lcr9875, the pretrained weights are only compatible with defalt models that does not use seperable convs. You may train your own models from scratch if you want to enable seperate convs.

from deeplabv3plus-pytorch.

@VainF Hi,I train it without pretrained weights by myself,but I get this problem while test the new weight.I read the code and i think the extra "body.x" in the weight is caused by -separable, so how should I test the weight with "body.x"? Thank you very much.

from deeplabv3plus-pytorch.