- Clone the InverseGraphics Julia package:
git clone [email protected]:probcomp/InverseGraphics.git
cd InverseGraphics
- Setup python3 virtual environment
python3 -m venv venv
source venv/bin/activate
pip install --upgrade pip setuptools
pip install -r requirements.txt
- Register the ProbComp registry in the Julia package manager:
registry add [email protected]:probcomp/InverseGraphicsRegistry.git
-
In the Julia REPL package manager run
activate .
and theninstantiate
-
Get the YCB-Video data. Download
- data (or get data_small if you just want a small subset of the full dataset)
- models
Extract the contents and place them in the data
such that the structure looks as follows:
data/
0048
...
0059
densefusion/
models/
model_list.txt
keyframe.txt
- Run
notebooks/demo.jl
in a jupyter/jupytext notebook. If all cells runs successfully, then the setup is functioning properly.
If you encounter issues with PyCall referencing the wrong python instance run: sh scripts/fix_pycall.sh
which runs the following command
PYTHON=$(which python) PYCALL_JL_RUNTIME_PYTHON=$(which python) julia --project -e 'import Pkg; Pkg.build("Conda"); Pkg.build("PyCall")'