Open hongsukchoi opened 2 years ago
I am using Ubuntu 20.04 and met some errors while building the environment.
Below are somewhat useful trouble-shooting tips.
Python3.6.9 conda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch
h5py numpy pyyaml==5.4.1 enum34 future scipy==1.4.1 opencv_contrib_python==3.4.2.16 matplotlib==3.0.2 transforms3d==0.3.1 scikit_image==0.13.1 lmdb==0.94 setuptools==41.0.0 cffi==1.11.5 easydict==1.7 plyfile==0.6 pillow==8.2.0 glumpy tqdm tensorboardX pandas scikit-learn scipy
find opencv / 3->4
pybind-11 https://github.com/ethnhe/FFB6D/issues/17#issuecomment-873929856
int -> cv::AccessFlag https://github.com/iterait/hipipe/issues/22#issuecomment-441410689
This env works well for all training/testing. But training with mixed precision dose not work.
I am using Ubuntu 20.04 and met some errors while building the environment.
Below are somewhat useful trouble-shooting tips.
Python / Pytorch
Python3.6.9 conda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch
New requirement
NormalSpeed
find opencv / 3->4
pybind-11 https://github.com/ethnhe/FFB6D/issues/17#issuecomment-873929856
int -> cv::AccessFlag https://github.com/iterait/hipipe/issues/22#issuecomment-441410689
This env works well for all training/testing. But training with mixed precision dose not work.