TeX-1500

TeX-1500 is a paired long-wave infrared hyperspectral imaging (LWIR HSI) dataset for temperature-emissivity-texture (TeX) decomposition. The associated paper describes 1,522 calibrated real-scene HSI-TeX pairs from DARPA pushbroom data and our FTIR acquisitions. Each full dataset sample pairs calibrated LWIR HSI with wavelength information, temperature, emissivity, and texture labels.

This repository is currently a preview release. It contains one example sample for format inspection and visualization. The full dataset will be made available in a later release.

Authors

• Cheng Dai (Hugging Face, GitHub)*
• Jiale Lin (Hugging Face, GitHub)*
• Hongyi Xu
• Bingxuan Song
• Ziyang Xie
• Fanglin Bao

*Equal Contribution.

Paper

• Hugging Face Papers: TeX-1500
• arXiv: 2606.03806

Preprocessing

The dataset preprocessing is based on HAIR:

• Paper: HADAR-Based Thermal Infrared Hyperspectral Image Restoration
• Code: jialelin2007/HAIR

Baseline

• Checkpoints: dccc2025/TeX-UNet
• Code: dccc2025/TeX-1500

Current Release Contents

.
├── README.md
├── LICENSE
├── NOTICE
├── CITATION.cff
├── data/
│   └── sample_0001/
│       ├── hsi.mat
│       ├── hsi_noisy.mat
│       ├── T.mat
│       ├── e.mat
│       ├── X.mat
│       └── previews/
│           ├── T.png
│           ├── e.png
│           ├── X.png
│           └── hsi_band.png
├── docs/
│   ├── DATA_FORMAT.md
│   └── RELEASE_STATUS.md
├── metadata/
│   ├── checksums.sha256
│   ├── dataset_summary.json
│   └── sample_manifest.jsonl
└── scripts/
    ├── inspect_sample.py
    └── verify_checksums.py

Example Sample

The included example is stored as MATLAB v5 .mat files:

• data/sample_0001/hsi.mat: denoised calibrated HSI, wavelengths, valid-band indices, sky estimates, and working wavelengths.
• data/sample_0001/hsi_noisy.mat: noisy input HSI.
• data/sample_0001/T.mat: temperature field T.
• data/sample_0001/e.mat: spectral emissivity field e.
• data/sample_0001/X.mat: texture field X.
• data/sample_0001/previews/: PNG previews for quick visual inspection.

Run the inspection helper with:

uv run python scripts/inspect_sample.py data/sample_0001

The helper prints each .mat variable name, shape, and dtype without loading the full arrays into memory.

Verify file integrity with:

uv run python scripts/verify_checksums.py

Data Format

See docs/DATA_FORMAT.md for the preview sample layout and variable schema. See metadata/sample_manifest.jsonl for machine-readable sample metadata.

Release Status

This is a documentation and one-sample preview release. The full 1,522-sample dataset described in the paper is not included yet. See docs/RELEASE_STATUS.md for the public-release boundary and what is intentionally excluded at this stage.

License

The dataset files and repository documentation are distributed under the PolyForm Noncommercial License 1.0.0. See LICENSE and NOTICE.

This license permits noncommercial use, including research and educational use, under its terms. Commercial use is not granted by this release.

Citation

If you use this dataset or preview sample, cite the TeX-1500 paper:

@misc{dai2026tex1500pairedrealworldlwir,
      title={TeX-1500: A Paired Real-World LWIR Hyperspectral Dataset and Benchmark for Temperature-Emissivity-Texture Decomposition}, 
      author={Cheng Dai and Jiale Lin and Hongyi Xu and Bingxuan Song and Ziyang Xie and Fanglin Bao},
      year={2026},
      eprint={2606.03806},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2606.03806}, 
}