DCNv2
DCNv2 copied to clipboard
CharlesShang
An error occurred when running ./make.sh
An error occurred when running ./make.sh ，Adding include_dirs.append('/usr/local/cuda/include')) still doesn't work
In file included from /home/---/.local/lib/python3.8/site-packages/torch/include/ATen/Tensor.h:3, from /home/---/.local/lib/python3.8/site-packages/torch/include/ATen/Context.h:4, from /home/---/.local/lib/python3.8/site-packages/torch/include/ATen/ATen.h:5, from /home/---/DCNv2-master/src/cpu/dcn_v2_cpu.cpp:4: /home/---/.local/lib/python3.8/site-packages/torch/include/ATen/core/TensorBody.h:354:7: note: declared here 354 | T * data() const { | ^~~~ /home/---/DCNv2-master/src/cpu/dcn_v2_cpu.cpp:224:9: error: ‘THFloatBlas_gemv’ was not declared in this scope; did you mean ‘THFloatBlas_gemm’? 224 | THFloatBlas_gemv('t', k_, m_, 1.0f, | ^~~~~~~~~~~~~~~~ | THFloatBlas_gemm error: command 'g++' failed with exit status 1
Did you found a solution @starsky68 ? I get the same error
Maybe your DCNv2 version is lower, please use DCNv2
I have tried this new DCNv2 which you provided, but still get the same error
Did you found a solution @starsky68 ? I get the same error
Maybe your DCNv2 version is lower, please use DCNv2
I have tried this new DCNv2 which you provided, but still get the same error
Did you found a solution @starsky68 ? I get the same error
Maybe your DCNv2 version is lower, please use DCNv2
https://github.com/lbin/DCNv2/tree/pytorch_1.6 @kkzh2313
I have tried this new DCNv2 which you provided, but still get the same error
Did you found a solution @starsky68 ? I get the same error
Maybe your DCNv2 version is lower, please use DCNv2
https://github.com/lbin/DCNv2/tree/pytorch_1.6 @kkzh2313
Is your user root or individual user, if it is an individual user, please add --user in make.sh
Just to be clear, if anyone will experience in the future this error, you should:
- delete the old version of DCNv2
- check which version of pytorch you have with
python3 -c "import torch; print(torch.__version__)" - clone the repository:
git clone https://github.com/lbin/DCNv2.git && cd DCNv2/ - checkout the correct branch. In my case:
git checkout pytorch_1.9
I hope this will help someone else in the same situation.
Is your user root or individual user, if it is an individual user, please add --user in make.sh
@starsky68, could you please be more precise about your statement?
It is not my case, but it would be helpful for people in this situation to know where they would we need to insert the --user option and how more precisely. Thanks!
This can compile successfully. (https://github.com/lbin/DCNv2/tree/pytorch_1.6)@lbin
And pytorch version is 1.6.0 (The installation command is "pip install torch==1.6.0 torchvision==0.7.0", where CUDA is 10.2)
