Datasetpublic

SPIN-Lab/KOREATECH-CGH-2048-3.6Mu

KOREATECH-CGH This dataset consists of RGBD–complex hologram pairs designed for training machine learning–based computer-generated holography (ML-CGH) models.It can be used for tasks such as hologram generation, hologram upscaling, and related applications. The holograms were generated using a layer-based hologram generation method[Article]. Note that this dataset is licensed under the Creative Commons Attribution 4.0 International License Non Commercial (CC BY-NC 4.0).… See the full description on the dataset page: https://huggingface.co/datasets/SPIN-Lab/KOREATECH-CGH-2048-3.6Mu.

sourceHugging Facecc-by-nc-4.0updated 3mo agoView on Hugging Face

1likes3.1kdownloads

Dataset Card

KOREATECH-CGH

This dataset consists of RGBD–complex hologram pairs designed for training machine learning–based computer-generated holography (ML-CGH) models. It can be used for tasks such as hologram generation, hologram upscaling, and related applications.

The holograms were generated using a layer-based hologram generation method[Article].

Note that this dataset is licensed under the Creative Commons Attribution 4.0 International License Non Commercial (CC BY-NC 4.0).

Dataset Sample

RGB	Depth
<img width="512" alt="RGB" src="rgb_2048.png">	<img width="512" alt="Depth" src="depth_2048.png">

Amplitude	Phase
<img width="512" alt="Amplitude" src="amplitude_2048.webp">	<img width="512" alt="Phase" src="phase_2048.webp">

Data Details

Directory structure

root
  ├─test
  │  ├─amp
  │     └─*.exr
  │  ├─depth
  │  ├─img
  │  └─phs
  ├─train
  │  ├─amp
  │  ├─depth
  │  ├─img
  │  └─phs
  └─validation
      ├─amp
      ├─depth
      ├─img
      └─phs

Dataset Configuration

	Format	Channels	Resolution	Precision	Range
RGB	.exr	3	2048 × 2048	fp32	0-1
Depth	.exr	1	2048 × 2048	fp32	0-1
Amplitude	.exr	3	2048 × 2048	fp32	dependent to data
Phase	.exr	3	2048 × 2048	fp32	0-1

Hologram Parameters

Parameter	Value
Resolution	2048 × 2048
Pixel Pitch	3.6 μm
Wavelength (R,G,B)	638 nm, 532 nm, 450 nm
Physical Extent (H × W × D)	7.3728 mm × 7.3728 mm × 81.33446 mm

Data Splits

Split	Number of Samples
Train	5,000
Validation	500
Test	500

Source 3D Models

The RGB-D scenes were generated from 3D meshes obtained from the Google Scanned Objects.

License

© 2025, SPIN Lab, Korea University of Technology and Education (KOREATECH) and Digital Holography Research Group, Electronics and Telecommunications Research Institute (ETRI) This dataset is licensed under the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) license. You are free to use, modify, and distribute this work for non-commercial purposes, with proper attribution. Commercial use is strictly prohibited.

See LICENSE and the official CC BY-NC 4.0 license for full terms.

For inquiries, please contact the corresponding author: bluekdct@gmail.com

Acknowledgements

This work was supported by the National Research Foundation of Korea (NRF) through the Ministry of Education's Basic Science Research Program (Grant 2021R1I1A3048263, 50\%) and by the Institute of Information and Communications Technology Planning and Evaluation (IITP) grant funded by the Korea Government (MSIT) (Grant 2019-0-00001, 50\%).

Citation

@article{LEE2026115636,
title = {A large-depth-range layer-based hologram dataset generation for machine learning-based 3D computer-generated holography},
journal = {Optics & Laser Technology},
volume = {203},
pages = {115636},
year = {2026},
issn = {0030-3992},
doi = {https://doi.org/10.1016/j.optlastec.2026.115636},
url = {https://www.sciencedirect.com/science/article/pii/S0030399226009874},
author = {Jaehong Lee and You Chan No and YoungWoo Kim and Duksu Kim},
keywords = {CGH(Computer-generated holography), Hologram, Machine-learning, Dataset, RGB-d, ML-CGH},
abstract = {Machine learning-based computer-generated holography (ML-CGH) has advanced rapidly in recent years, yet progress is constrained by the limited availability of high-quality, large-scale hologram datasets. To address this, we present KOREATECH-CGH, a publicly available dataset comprising 6000 pairs of RGB-D images and complex holograms across resolutions ranging from 256×256 to 2048×2048, with depth ranges extending to the theoretical limits of the angular spectrum method for wide 3D scene coverage. To improve hologram quality at large depth ranges, we introduce amplitude projection, a post-processing technique that replaces amplitude components of hologram wavefields at each depth layer while preserving phase. This approach enhances reconstruction fidelity, achieving 27.46 dB PSNR and 0.87 SSIM, surpassing a recent optimized silhouette-masking layer-based method by 3.86 dB and 0.09 SSIM, respectively. We further validate the utility of KOREATECH-CGH through experiments on hologram generation and super-resolution using state-of-the-art ML models, confirming its applicability for training and evaluating next-generation ML-CGH systems.}
}