Closed gmt710 closed 3 years ago
@gmt710 Have you installed vedadet as following? pip install -r requirements/build.txt pip install -v -e .
Yeah, I'll try again.
Can I use torch==1.1.0 ? Because my cuda version is 9.0.
@gmt710 we recommend using torch==1.6.0 or higher.
@hxcai ,thanks for your reply.My enviroment ,cuda9.2 torch1.6.0 test good. I have another problem about how to prepare widerface datasets for train and test ? Refers the code comments and the bugs, I prepare widerface data but test and train are not working properly.
Thank you for your sharing. But I have this issue,too. I don't know how to figure it out.
ImportError: cannot import name 'deform_conv_ext'
I have try this : pip install -r requirements/build.txt pip install -v -e .
But I've got :
Using pip 20.2.4 from /home/cv/anaconda3/envs/tinaface/lib/python3.7/site-packages/pip (python 3.7)
Non-user install because site-packages writeable
Created temporary directory: /tmp/pip-ephem-wheel-cache-l6jhmr6r
Created temporary directory: /tmp/pip-req-tracker-mxmsm8pp
Initialized build tracking at /tmp/pip-req-tracker-mxmsm8pp
Created build tracker: /tmp/pip-req-tracker-mxmsm8pp
Entered build tracker: /tmp/pip-req-tracker-mxmsm8pp
Created temporary directory: /tmp/pip-install-yi04rbse
Looking in indexes: https://pypi.tuna.tsinghua.edu.cn/simple
Obtaining file:///home/cv/PycharmProjects/vedadet
Added file:///home/cv/PycharmProjects/vedadet to build tracker '/tmp/pip-req-tracker-mxmsm8pp'
Running setup.py (path:/home/cv/PycharmProjects/vedadet/setup.py) egg_info for package from file:///home/cv/PycharmProjects/vedadet
Created temporary directory: /tmp/pip-pip-egg-info-63ncdptk
Running command python setup.py egg_info
/home/cv/anaconda3/envs/tinaface/lib/python3.7/site-packages/torch/cuda/init.py:52: UserWarning: CUDA initialization: The NVIDIA driver on your system is too old (found version 10010). Please update your GPU driver by downloading and installing a new version from the URL: http://www.nvidia.com/Download/index.aspx Alternatively, go to: https://pytorch.org to install a PyTorch version that has been compiled with your version of the CUDA driver. (Triggered internally at /opt/conda/conda-bld/pytorch_1603729138878/work/c10/cuda/CUDAFunctions.cpp:100.)
return torch._C._cuda_getDeviceCount() > 0
Traceback (most recent call last):
File "
@Kerr46 It seems that your environment has no cuda?
@hxcai , Greatly Thanks. I'll try.
@hxcai ,I follow this ,also met same problem. Maybe parse xml do not work.
@gmt710 please check in #11
@mike112223 ,Thanks. I'll try.
I took some notes for reference. Anyone needed can click this.(more informations)
my enviroment:
ubuntu16.04
python3.6
cuda9.2
pytorch 1.6.0
Now the reference include preparing enviroment, datasets and testing on widerface val datasets. If I have time, I will update it.
@gmt710 Great, thanks for sharing!
Thanks. The result is same with your paper. I test no TTA(the result is good):
If I want to use cuda9.0, Do you have any suggestions for me to modify dcn? I've tried the version of dcn from mmdetection several times but fail .
@gmt710 You can try to use dcn in old version mmdetection that supports cuda9.0. But we don't recommend to do that.
@hxcai , Is this mainly because there is a corresponding relationship between torch and CUDA versions?
@gmt710 specified torch and dcn implementation is related to specified cuda version.
I am still facing the same problem after doing all of what was suggested above, can anyone help?
Greatly thank for your sharing. when I run the command:
I met this problem:
Maybe this cause by dcn, Would you mind giving some suggestions for me? Thx.