Installation is a nightmare

[openSourcerer9000]

conda create -n fvid python=3.7; conda activate fvid; conda install pytorch==1.7.1 torchvision==0.8.2 cudatoolkit=10.1; conda install ffmpeg==4.0.2; pip install opencv-python dominate scipy tqdm matplotlib scikit-image;

install.sh

-CUDA HOME env var error-

conda install -c conda-forge cudatoolkit-dev

subprocess.CalledProcessError: Command '['which', 'g++']' returned non-zero exit status 1.

sudo apt install build-essential

subprocess.CalledProcessError: Command '['ninja', '-v']' returned non-zero exit status 1.

apparently this goes back to the initial cuda home issue I tried ./launch_docker.sh instead. Should be easier with docker right?

sudo: nvidia-docker: command not found

Apparently that's deprecated, need to install a bunch of extra stuff. https://github.com/NVIDIA/nvidia-docker#quickstart https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html

E: Release file for http://archive.ubuntu.com/ubuntu/dists/jammy-updates/InRelease is not valid yet (invalid for another 3h 4min 23s). Updates for this repository will not be applied.

Has ANYONE ever managed to install this?

[hacker009-sudo]

exactly impossible to install it seems :(

[Varun0157]

A bit late to the party, but give this a look. I managed to install flownet2-pytorch locally in a conda environment by following the steps linked.

[caiyinfan]

Hi,your porblem solved yet?can you give me some help to settle install flownet2-pytorch，thank you