QRF Dataset
- Citation Author(s):
-
YuanFu Yang
(National Yang Ming Chiao Tung University)
- Submitted by:
- Yang YuanFu
- Last updated:
- DOI:
- 10.21227/6eke-rb86
- Data Format:
12 views
- Categories:
- Keywords:
Abstract
The QRF dataset is designed to support research in quantum-native photorealistic scene rendering. It consists of high-fidelity 3D indoor and outdoor environments captured from multiple calibrated viewpoints, with detailed annotations of geometry, material properties, and lighting conditions. Each scene is processed into quantum-compatible representations for training and evaluating Quantum Radiance Fields (QRF), which leverage quantum circuits, activation functions, and quantum volume rendering.
This dataset serves as a benchmark for evaluating the efficiency and realism of quantum implicit representations. It includes both low-frequency and high-frequency features to challenge classical rendering models and demonstrate the expressive power of quantum processing. The dataset facilitates experimentation in tasks such as view synthesis, radiance prediction, and ray-based integration under a quantum computing paradigm. By providing carefully curated and diverse scene data, the QRF dataset enables rigorous testing of quantum advantages in 3D vision and computational graphics.
Instructions:
QRF Dataset Instructions
The QRF Dataset contains 8 diverse scenes (Bike, Lifestyle, Palace, Robot, Spaceship, Steamtrain, Toad, Wineholder) for developing and evaluating quantum-native 3D rendering and reconstruction methods.
Each scene is organized with the following structure:
<SceneName>/
├── pose/ # Camera extrinsic matrices (one file per view)
├── rgb/ # RGB images of the scene
├── bbox.txt # Axis-aligned bounding box of the scene
└── intrinsics.txt # Camera intrinsic matrix
Camera Parameters
intrinsics.txt:
Stores the 3×3 camera intrinsic matrix, formatted as:fx 0 cx
0 fy cy
0 0 1
pose/:
Each file contains a 4×4 camera-to-world transformation matrix per view:[ R | t ]
[ 0 0 0 1 ]
RGB Images
- Stored in
rgb/folder. - Filenames correspond to pose files (e.g.,
0001.png↔0001.txtinpose/).
Bounding Box
bbox.txt: Defines the scene’s 3D bounding region for rendering or normalization.xmin xmaxymin ymaxzmin zmax
Usage
This dataset is intended for training Quantum Radiance Fields (QRF) or other implicit neural representations under quantum or classical rendering pipelines. The structured and calibrated scenes support:
- View synthesis
- Neural rendering
- Ray-based integration
- Scene reconstruction
Citation
If you use this dataset in your research, please cite:
@article{yang2025qrf,
title={Quantum Radiance Fields: A Fully Quantum-Native Framework for Photorealistic Neural Rendering},
author={Yang, YuanFu et al.},
journal={TPAMI},
year={2025}
}
