Viktor Rudnev, Mohamed Elgharib, Christian Theobalt, Vladislav Golyanik

Based on NeRF-OSR codebase, which is based on NeRF++ codebase and inherits the same training data preprocessing and format.

Download the datasets from here.

Untar the downloaded archive into data/ sub-folder in the code directory.

See NeRF++ sections on data and COLMAP on how to create adapt a new dataset for training.

Please contact us if you need to adapt your own event stream as it might need updates to the code.

conda env create --file environment.yml
conda activate eventnerf

Use the scripts from scripts/ subfolder for training and testing. Please replace <absolute-path-to-code> and <path-to-conda-env> in the .sh scripts and the corresponding .txt config file To do so automatically for all of the files, you can use sed:

sed 's/<absolute-path-to-code>/\/your\/path/' configs/**/*.txt scripts/*.sh
sed 's/<path-to-conda-env>/\/your\/path/' scripts/*.sh

• configs/nerf/*, configs/lego1/* -- synthetic data,
• configs/nextgen/*, configs/nextnextgen/* -- real data (from the revised paper),
• configs/ablation/* -- ablation studies,
• configs/altbase.txt -- constant window length baseline,
• configs/angle/* -- camera angle error robustness ablation,
• configs/noise/* -- noise events robustness ablation,
• configs/deff/* -- data efficiency ablation (varying amount of data by varying the simulated event threshold),
• configs/e2vid/* -- synthetic data e2vid baseline,
• configs/real/* -- real data (from the old version of the paper)

To extract the mesh from a trained model, run

ddp_mesh_nerf.py --config nerf/chair.txt

Replace nerf/chair.txt with the path to your trained model config.

Please find the guide on evaluation, color-correction, and computing the metrics in metric/README.md.

Please cite our work if you use the code.

@InProceedings{rudnev2023eventnerf,
      title={EventNeRF: Neural Radiance Fields from a Single Colour Event Camera},
      author={Viktor Rudnev and Mohamed Elgharib and Christian Theobalt and Vladislav Golyanik},
      booktitle={Computer Vision and Pattern Recognition (CVPR)},
      year={2023}
}

This work is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-sa/4.0/ or send a letter to Creative Commons, PO Box 1866, Mountain View, CA 94042, USA.