bumbledeep/fissl

Dataset Card for Dataset Name

Fundus Images for Self Supervised Learning (FISSL)

Dataset Details

Dataset Description

FISSL is a dataset resulting from the combination of 4 different retinal image datasets (RFMID, ODIR, eyePACS, APTOS). All images have been resized to 224x224 and converted to .png format

• Curated by: Diego Hernández
• License: MIT

Dataset Sources

• Repository:
  • huggingface dataset RFMID
  • huggingface dataset ODIR
  • huggingface dataset eyePACS
  • huggingface dataset APTOS

Uses

Direct Use

This version of the dataset is unlabeled and intended for feature extraction. More specifically, it was designed with self-supervised learning in mind.

Dataset Structure

Dataset Creation

Curation Rationale

The images have not been edited except for resizing. No additional processing has been performed.

Source Data

The used sources are properly described in each dataset card of the list given in the Dataset Sources section.

Data Collection and Processing

Composition of the dataset (original sources):

• RFMID
• ODIR
• eyePACS
• APTOS

Who are the source data producers?

See the dataset cards of each source.

Bias, Risks, and Limitations

The heterogeneity in the images can be quite high due to the lack of a rigorous standardization process. This may affect the quality of the analyses performed.

Recommendations

It is recommended to preprocess the images before using them to train deep learning models.

Citation

If you use this dataset, please consider mentioning this repository or my Hugging Face username :)

Dataset Card Authors

bumbledeep