conda create -n panogrf python=3.7
conda activate panogrf
# pip install pytorch for CUDA11.1
pip install torch==1.10.1+cu111 torchvision==0.11.2+cu111 torchaudio==0.10.1 -f https://download.pytorch.org/whl/cu111/torch_stable.html
# requirements
cd PanoGRF
pip install -r requirements.txt
Make sure you have installed EGL + OpenGL in your environment. Check out by egl-example.
If you haven't installed egl, egl installation procedure from docker file is like this.
use pip:
pip install git+https://github.com/facebookresearch/habitat-lab.git@v0.2.2
If you fail to install habitat-lab when using pip, you can also install it from source code. Refer to official habitat repository
conda install habitat-sim=0.2.2 headless -c conda-forge -c aihabitat -y
This step takes a long time for me.
or install from local file.
# download `habitat-sim-0.2.2-py3.7_headless_bullet_linux_011191f65f37587f5a5452a93d840b5684593a00.tar.bz2` from `https://anaconda.org/aihabitat/habitat-sim/files`
conda install --use-local habitat-sim-0.2.2-py3.7_headless_bullet_linux_011191f65f37587f5a5452a93d840b5684593a00.tar.bz2
if your habitat environment cannot work, you may check your environment via requirements_freeze.txt
...
-e git+https://github.com/facebookresearch/habitat-lab.git@0f454f62e41050bc90ca468c62db35d7484923ff#egg=habitat
-e git+https://github.com/facebookresearch/habitat-lab.git@afe4058a7f8aa5ab71a133575cdaa79f0308af6a#egg=habitat_lab&subdirectory=habitat-lab
habitat-sim==0.2.2
...