saimouli/LiDiff

Scaling Diffusion Models to Real-World 3D LiDAR Scene Completion

Installing python packages pre-requisites:

sudo apt install build-essential python3-dev libopenblas-dev

pip3 install -r requirements.txt

Installing MinkowskiEngine:

pip3 install -U MinkowskiEngine==0.5.4 --install-option="--blas=openblas" -v --no-deps

To setup the code run the following command on the code main directory:

pip3 install -U -e .

Diffusion Scene Completion Pipeline

For training the diffusion model, the configurations are defined in config/config.yaml, and the training can be started with:

python3 train.py

For training the refinement network, the configurations are defined in config/config_refine.yaml, and the training can be started with:

python3 train_refine.py

For running the scene completion inference we provide a pipeline where both the diffusion and refinement network are loaded and used to complete the scene from an input scan. You can run the pipeline with the command:

python3 tools/diff_completion_pipeline.py --diff DIFF_CKPT --refine REFINE_CKPT -T DENOISING_STEPS -s CONDITIONING_WEIGHT