Skip to content

Latest commit

 

History

History
44 lines (36 loc) · 1.78 KB

README.md

File metadata and controls

44 lines (36 loc) · 1.78 KB

Low Cost Single Photon Camera Dataset

Dataset of measurements from a low-cost single-photon camera used in our CVPR 2024 paper "Towards 3D Vision with Low-Cost Single-Photon Cameras"

(Project Webpage | PDF)

teaser image

Dataset

The dataset contains both real and simulated data. Real data contains posed captures from the AMS TMF8820. Simulated data is from our own simulator. Real-world captures were completed by attaching the sensor to a robot arm, which moves around the object while capturing measurements.

Not all real-world captures have ground truth. Those that do contain a .stl file and a .blend file of the ground truth mesh.

File Format

Data is stored in .json files with the following format:

[
    {
        "reference_hist": [...], // (128,) array (real data only)
        "hists": [...], // (9, 128) array (real data) or (128,) array (simulated data)
        "pose": [...] // (4, 4) homogeneous transform from origin to sensor pose
    },
    {
        // repeat for as many measurements
    }
]

For real data, there are additional fields (e.g. sensor temperature, depth estimates, robot state). These can be ignored.

Troubleshooting

If you run into trouble, please make an issue on this GitHub repo.

Citation

If you use this dataset for published work, please cite our CVPR 2024 paper:

@article{Mu24Towards3DVision,
    title={Towards 3D Vision with Low-Cost Single-Photon Cameras}, 
    author={Fangzhou Mu and Carter Sifferman and Sacha Jungerman and Yiquan Li and Mark Han and Michael Gleicher and Mohit Gupta and Yin Li},
    year={2024},
    eprint={2403.17801},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}