l11p/nffa-europe-sem-dataset

NFFA-EUROPE Majority SEM Dataset

A Hugging Face dataset containing 25,537 Scanning Electron Microscopy (SEM) images classified into 10 categories. Originally published by the NFFA-EUROPE project, this dataset is suitable for computer vision research in materials science and microscopy analysis.

Dataset Description

Total Images: 25,537 SEM images
Classes: 10 categories
Format: Mostly JPEG images (with possible PNG/JPEG variants), various resolutions

Categories

The dataset contains images from the following categories:

• Biological (972 images)
• Fibres (162 images)
• Films_Coated_Surface (326 images)
• MEMS_devices_and_electrodes (4,688 images)
• Nanowires (3,820 images)
• Particles (3,925 images)
• Patterned_surface (8,922 images)
• Porous_Sponge (181 images)
• Powder (917 images)
• Tips (1,624 images)

Data Source

This dataset is sourced from the NFFA-EUROPE Majority SEM Dataset, originally published and made available through the B2SHARE federated research data repository maintained by EUDAT.

Citation

If you use this dataset in your research, please cite the original work:

@dataset{aversa2018nffa,
  title={NFFA-EUROPE - Majority SEM Dataset},
  author={Aversa, Rossella and Modarres, Mohammad Hadi and Cozzini, Stefano and Ciancio, Regina},
  howpublished={B2SHARE},
  year={2018},
  month={February},
  day={19},
  url={https://b2share.eudat.eu/records/1nysp-05236},
  doi={10.23728/b2share.1nysp05236},
  note={NFFA-EUROPE project, Version 1.0}
}

Data Source Details

• Repository: B2SHARE (https://b2share.eudat.eu)
• Record ID: 1nysp-05236
• Original Landing Page: https://b2share.eudat.eu/records/1nysp-05236
• Publication Date: February 19, 2018
• Version: 1.0

Original Dataset Description

From the original publication:

Dataset of 25,537 SEM images produced at CNR-IOM (Trieste, Italy). Images are classified into 10 categories in a folder structure. Classification labels have been checked by a group of microscientists on the web site http://sem-classifier.html and only those images which have been validated by the absolute majority of the group have been included in the dataset. The dataset is appropriate for the purposes of this study and in general for visual object recognition software research. Any scientific metadata associated with the measure is not present in the images. The dataset is therefore relevant as a whole, being the single images entirely detached from any specific information or scientific detail related to the displayed subject. This work has been done within the NFFA-EUROPE project (www.nffa.eu) and has received funding from the European Union's Horizon 2020 Research and Innovation Programme under grant agreement No. 654360 NFFA-Europe.

Dataset Structure

The dataset is organized within Hugging Face as follows:

Dataset columns:
├── image (Image): image binary payload (lazy decoded)
├── label (string): Category label (folder name)
└── image_filename (string): Original filename

Usage

Load the dataset from Hugging Face:

from datasets import load_dataset

# Load the full dataset
dataset = load_dataset("l11p/nffa-europe-sem-dataset")

# Access a single example
example = dataset["train"][0]
print(example.keys())  # dict_keys(['image', 'label', 'image_filename'])

# Access image and label
image = example["image"]
label = example["label"]
filename = example["image_filename"]

# Iterate over examples
for sample in dataset["train"]:
    image = sample["image"]
    label = sample["label"]
    # Process...

License

The dataset is released under the terms of the original NFFA-EUROPE project. See the original B2SHARE record for complete license information: https://b2share.eudat.eu/records/1nysp-05236

Funding

This dataset was funded by the European Union's Horizon 2020 Research and Innovation Programme under grant agreement No. 654360 NFFA-Europe.

Technical Details

• Source Repository: B2SHARE federated research data repository
• Download Method: Automated via B2SHARE API
• Processing: Images extracted from tar archives and organized into Hugging Face datasets format
• Image Encoding: Lazy-loaded via Hugging Face Image feature (decode=False)
• Source Archive Size: Approximately 16.8 GB (10 B2SHARE tar archives)
• Published Dataset Size (train.num_bytes): ~18.5 GB

Related Work

For information about the NFFA-EUROPE project, visit: https://www.nffa.eu/

Acknowledgments

We acknowledge the original authors and CNR-IOM (Trieste, Italy) for producing and publicly sharing this valuable dataset through the B2SHARE repository and the NFFA-EUROPE project.

---

This dataset was packaged for Hugging Face from the original B2SHARE source. For detailed metadata and download options, visit the original record at https://b2share.eudat.eu/records/1nysp-05236