April 7, 2025 Dataset Open Access

Electrolyzers-HSI: Close-Range Multi-Scene Hyperspectral Imaging Benchmark Dataset

Arbash, Elias; Afifi, Ahmed Jamal; Belahsen, Ymane; Fuchs, Margret; Ghamisi, Pedram; Scheunders, Paul; Gloaguen, Richard

Electrolyzers-HSI Dataset

Description:

The Electrolyzers-HSI dataset is a multiscene RGB-Hyperspectral benchmark dataset comprising 55 scene of shredded Electrolyzers samples. The RGB images are collected using a Teledyne Dalsa C4020 camera on a conveyor belt, while hyperspectral images (HSI) are acquired with a FENIX spectrometer. The HSI data contains 450 bands in the VNIR and SWIR range [400 - 2500]nm.

Data Format

• RGB Images: .jpg files
• Ground Truth (GT): .png files. They appear black since the values are between 0 and 5. Correct visualization is done via script.
• HSI Data: Each hyperspectral data cube .img file is accompanied by a .hdr file.

Folder Organization

• Electrolyzers-HSI: 55 subfolders 
  • 1/
    • ’GT.png’ file for segmentation ground truth
    • ‘HSI.img’ and ‘HSI.hdr’ files for HSI data cube
    • ‘RGB.jpg’ file for the RGB image
  • 2/
    • ’GT.png’ file for segmentation ground truth
    • ‘HSI.img’ and ‘HSI.hdr’ files for HSI data cube
    • ‘RGB.jpg’ file for the RGB image
  • 3/4/5/6/ … :Same structure for all rest of folders

Data Classes in Masks

• Masks contain 0 to 5 segmentation classes:
  • 0: background
  • 1: “MESH”
  • 2: “Steel_Cathode”
  • 3: "Steel_Anode”
  • 4: “HTEL_Anode”
  • 5: “HTEL_Cathode”

Code Repository

To facilitate reading and working with the data, Python codes are available on the GitHub repository:

https://github.com/hifexplo

Citation

If you use this dataset, please cite the following article:

Word:

Latex: