Awesome

Domain Reduction Strategy for Non-Line-of-Sight Imaging

Hyunbo Shim^*, In Cho^*, Daekyu Kwon, Seon Joo Kim (* Equal contribution)

[arXiv] [BibTeX]

An official implementation of the paper Domain Reduction Strategy for Non-Line-of-Sight Imaging (in ECCV 2024).

Preparation

Starting with docker

We provide a prebuilt docker image with the required packages installed.

docker pull join16/join16/nlos-domain-reduction:py39-cu113

Or you can also build a docker image by running the following command.

docker build -t nlos-domain-reduction:py39-cu113 .

Installing packages with pip

Instead of using a docker image, you can also manually install the required packages using pip. We recommend using a virtual environment to avoid conflicts with other packages.

pip install -r requirements.txt

We tested our code on Python 3.9, torch 2.0.1, and CUDA 11.3.

Building CUDA kernel

Our forward propagation model (hidden scenes to measurements) is implemented as a CUDA kernel. To build the CUDA kernel, run the following command.

python setup.py build_ext --inplace

Logging with wandb (optional)

To log experiment results using wandb, place the wandb API key in the config/wandb.yaml file.

project: nlos-domain-reduction
entity: {your_wandb_username}
api_key: {your_wandb_api_key}

Dataset

For the ZNLOS dataset, first download the original data from the official website. Then, modify raw_root_dir in the config/data/znlos.yaml to the root directory of the downloaded data. For the Stanford real world dataset, download the original data and modify raw_root_dir in the config/data/stanford.yaml to the path of the downloaded data. Our evaluation script will automatically preprocess the data.

Reconstruction

Run our model using the following command.

python main.py config/main_bunny.yaml -n {experiment_name} -g {gpu_id}

Acknowledgements

We sincerely appreciate the authors for sharing their code and data, which greatly helped our research.

<a name="citation"></a> Citation

@article{shim2024drs,
  author    = {Shim, Hyunbo and Cho, In and Daekyu, Kwon and Kim, Seon Joo},
  title     = {Domain Reduction Strategy for Non Line of Sight Imaging},
  journal   = {Proceedings of the European Conference on Computer Vision (ECCV)},
  year      = {2024},
}