EDVR
EDVR copied to clipboard
Problem with importing ModulatedDeformConvPack
Hi Everyone, I am trying to train the EDVR in google colab with torch.version '1.3.0+cu100' I have compliled the DCN module first: (!python setup.py develop) and this is the result:
running develop running egg_info writing deform_conv.egg-info/PKG-INFO writing dependency_links to deform_conv.egg-info/dependency_links.txt writing top-level names to deform_conv.egg-info/top_level.txt reading manifest file 'deform_conv.egg-info/SOURCES.txt' writing manifest file 'deform_conv.egg-info/SOURCES.txt' running build_ext copying build/lib.linux-x86_64-3.6/deform_conv_cuda.cpython-36m-x86_64-linux-gnu.so -> Creating /usr/local/lib/python3.6/dist-packages/deform-conv.egg-link (link to .) Adding deform-conv 0.0.0 to easy-install.pth file Installed /content/gdrive/My Drive/Colab Notebooks/EDVR/EDVR-master/codes/models/archs/dcn Processing dependencies for deform-conv==0.0.0 Finished processing dependencies for deform-conv==0.0.0
However, when I try to import ModulatedDeformConvPack as DCN, I face ImportError in this line: from models.archs.dcn import deform_conv_cuda
ImportError: /content/gdrive/My Drive/Colab Notebooks/EDVR/EDVR-master/codes/models/archs/dcn/deform_conv_cuda.cpython-36m-x86_64-linux-gnu.so: undefined symbol: _ZN3c1011CPUTensorIdEv
Could anybody help me to solve this problem?
I have the same question
How to solve it?
i am so sad to meet the same question with you
I meet the same problem. My suggestion is to look for answers in https://github.com/open-mmlab/mmdetection/issues/1577. Still trying,
In my case, I removed all previous build files from /archs/dcn folder including builds, deform_conv.egg-info , pycache, and .so file, then I rebuilt them by running python setup.py develop
, this solved the problem.
In my case, I removed all previous build files from /archs/dcn folder including builds, deform_conv.egg-info , pycache, and .so file, then I rebuilt them by running
python setup.py develop
, this solved the problem.
Absolutely right.