Balanced Spherical Grid for Egocentric View Synthesis

Official implementation of Balanced Spherical Grid for Egocentric View Synthesis (CVPR 2023)

Project Page | Paper | Data

TL;DR - One Sentence Summary

EgoNeRF is a grid-based NeRF model that utilize a balanced spherical grid to reconstruct large-scale egocentric captured scenes.

---

Installation

Environment Setup

We tested our code on Ubuntu 20.04 with RTX 3090 GPU. With proper version of CUDA toolkit, it would work on other environments.

conda create -n EgoNeRF python=3.8
conda activate EgoNeRF
pip install torch torchvision --extra-index-url https://download.pytorch.org/whl/cu113
pip install -r requirements.txt

Dataset

We provide OmniBlender and Ricoh360 datset. You can download the dataset from the google drive link above. Put the data in the directory data/

Training EgoNeRF

To train EgoNeRF, run the scripts below.

python train.py --config configs/omniblender/barbershop/default.txt

Citation

Cite as below if you find this paper and repository are helpful to you:

@InProceedings{Choi_2023_CVPR,
    author    = {Choi, Changwoon and Kim, Sang Min and Kim, Young Min},
    title     = {Balanced Spherical Grid for Egocentric View Synthesis},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2023},
    pages     = {16590-16599}
}

Acknowledgement

This repository borrows codes from the amazing work TensoRF.