You need to agree to share your Hugging Face information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Access requests are reviewed manually. Every sample carries its own Creative Commons license (field license_code) and an attribution string that must be preserved when the material is reused.

Log in or Sign Up to review the conditions and access this dataset content.

LAION Cell 000

Part 000 of the laion/laion-cell-NNN series: figure images and supplementary videos harvested from the PubMed Central (PMC) Open Access subset, each with its caption, the sentences of the article that cite it, article metadata and the article's license. Repos of the series are filled one after another (~1000 GB each).

The series is produced by a continuous harvester that processes every OA article version listed in NCBI's PMC Article Datasets on AWS (s3://pmc-oa-opendata): it reads the JATS XML, resolves every captioned <fig> graphic and every video <media> / <supplementary-material>, and stores the original files byte-for-byte (md5-verified against PMC). Articles likely to contain cell imagery and cell videos were processed first.

Contents

config files sample
images images/NNNNNN.tar (WebDataset) <key>.jpg/.webp/.png/.gif/.tif + <key>.json
videos videos/NNNNNN.tar (WebDataset) <key>.mp4/.avi/.mov/... + <key>.json
metadata `metadata/<images videos>-NNNNNN.parquet`

Selected fields of <key>.json / the metadata parquet:

  • caption, caption_title, label — figure / video legend as published; mentions — up to 3 sentences of the article body that cite the item
  • cell_match, score, terms, modalities — pre-tags from a transparent caption keyword scorer (cell_match = 1: the caption describes microscopy / cells / tissue; modalities e.g. fluorescence, electron, histology, brightfield, live, multiphoton, superres, lightsheet, ish). Not a manual annotation.
  • width, height, frames, duration, fps, codec — media properties; md5, bytes, url (source object)
  • pmcid, pmid, doi, title, journal, year, authors, keywords, subjects, abstract, article_url
  • license_code, license_url, copyright, attribution
from datasets import load_dataset
ds = load_dataset("laion/laion-cell-000", "images", split="train", streaming=True)
meta = load_dataset("laion/laion-cell-000", "metadata", split="train")   # captions/licenses only

Licensing

Only articles whose PMC license_code is CC0, CC BY, CC BY-SA, CC BY-ND, CC BY-NC, CC BY-NC-SA or CC BY-NC-ND are included; author manuscripts (text-mining only) and articles with custom licenses are excluded. The license is per sample — filter on license_code for your use (e.g. only CC0 / CC BY / CC BY-SA for commercial use, no modification of *-ND items when redistributing). Keep the attribution field with any reuse. Some figures in CC-licensed articles reproduce third-party material (captions often say "reprinted/adapted with permission"); such content may carry its own restrictions.

Source and terms

Data from the NIH NLM NCBI PubMed Central (PMC) Article Datasets, accessed via https://registry.opendata.aws/ncbi-pmc. NLM is acknowledged as the source of the data; NLM/NIH/HHS do not endorse this dataset. The dataset reflects PMC at harvest time and may not reflect the most current or accurate data available from NLM (articles get corrected, updated or retracted; retracted articles were excluded at harvest time). Each sample links back to its source article.

Downloads last month
637