SPECTRE Dataset

This repository contains the dataset and code for the paper "SPECTRE: A Dataset for Spectral Reconstruction on Chip-Size Spectrometers With a Physics-Informed Augmentation Method". You can find all data files in the Zenodo repository

Contents

• examples/: Contains examples on how to use the dataset.
• src/: Contains the python files of the spectre module.

Requirements

• Python 3.11+
• Dependencies are listed in the pyproject.toml file.

Usage

1. Clone the repository:

   git clone https://github.com/your-repo.git

2. Install the project:

   pip install .

3. Use the example scripts or notebooks:

   python examples/read_data.py

License

• The dataset is available under the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) License.
• The code is available under the Software Copyright License for Academic Use of the Fraunhofer Software, Version 1.0.

Citation

If you use this dataset or code in your work, please cite the corresponding paper:

SPECTRE: A Dataset for Spectral Reconstruction on Chip-Size Spectrometers With a Physics-Informed Augmentation Method

Available in the Proceedings of IEEE Sensors 2024. You can access the paper via IEEE Xplore

BibTeX entry:

@INPROCEEDINGS{10784898,
  author={Wissing, Julio and Scholz, Teresa and Saloman, Stefan and Fargueta, Lidia and Junger, Stephan and Stefani, Alessio and Tschekalinskij, Wladimir and Scheele, Stephan and Schmid, Ute},
  booktitle={2024 IEEE SENSORS}, 
  title={SPECTRE: A Dataset for Spectral Reconstruction on Chip-Size Spectrometers with a Physics-Informed Augmentation Method}, 
  year={2024},
  volume={},
  number={},
  pages={1-4},
  keywords={Optical filters;Training;Neural networks;Reconstruction algorithms;Benchmark testing;Data augmentation;Optical sensors;Intelligent sensors;Optical arrays;Optical Sensors;Machine Learning;Artificial Intelligence;Spectral Reconstruction;Data Augmentation},
  doi={10.1109/SENSORS60989.2024.10784898}}