This repo contains the projects: 'Virtual Normal', 'DiverseDepth', and '3D Scene Shape'. They aim to solve the monocular depth estimation, 3D scene reconstruction from single image problems.
The code as written doesn't work w/ my RTX 3090. Not supported capability. I built a docker (that works!) -
docker pull petergaston/u18cuda11
This is:
base nvcr.io/nvidia/cuda:11.0.3-cudnn8-runtime
installation per instructions, except for line:
conda install pytorch==1.7.0 torchvision==0.8.0 torchaudio==0.7.0 cudatoolkit=11.0 -c pytorch
think I also needed these to get it to work...
apt-get update
apt install libgl1-mesa-glx # some error...
apt-get install -y libglib2.0-0 libsm6 libxrender1 libxext6
The code as written doesn't work w/ my RTX 3090. Not supported capability. I built a docker (that works!) -
docker pull petergaston/u18cuda11
This is: