Welcome to the official codebase for the Neural Kaleidoscopic Space Sculpting paper, a project that presents a novel method for single-shot full-surround 3D reconstruction using a kaleidoscopic image.
"Neural Kaleidoscopic Space Sculpting"
Byeongjoo Ahn, Michael De Zeeuw, Ioannis Gkioulekas, and Aswin C. Sankaranarayanan
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023
This repository is based on the IDR repository, which serves as the foundation for our work.
Create and activate the conda environment using the provided environment.yml file. Run the following commands in your terminal:
conda env create -f environment.yml
conda activate nkss
You can download the required kaleidoscopic image and calibration data here. Please ensure to place it in the data/
directory.
To initiate the training process, navigate to the code directory and run the following commands:
python training/exp_runner.py --conf ./confs/toy.conf
Post-training, you can extract the meshed surface by executing the commands below:
python evaluation/eval.py --conf ./confs/toy.conf --resolution 200 --eval_levelset --scale 0.25 --eval_rendering
Our group has also explored other areas in imaging using mirrors. You might find these related works interesting:
- Kaleidoscopic Structured Light (TOG 2021)
- Wide-Baseline Light Fields using Ellipsoidal Mirrors (PAMI 2022)
If you find our work useful or inspiring, please consider citing our paper using the following Bibtex entry:
@inproceedings{ahn2023neural,
author = {Ahn, Byeongjoo and De Zeeuw, Michael and Gkioulekas, Ioannis and Sankaranarayanan, Aswin C.},
title = {Neural Kaleidoscopic Space Sculpting},
booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
year = {2023}
}