Skip to main content

Cataract-LMM: Large-Scale Multi-Source Multi-Task Cataract Surgery Benchmark

3,000 cataract surgery procedures, 1,134.2 hours of video with phase annotations, instance segmentation, tracking, and skill scoring. Largest public cataract-surgery-video resource.

At a glance

FieldValue
Short namecataract_lmm
Full nameCataract-LMM: Large-Scale Multi-Source Multi-Task Cataract Surgery Benchmark
First published2026
Publication date precisionyear
Publication date evidenceapi.datacite.org/dois
Publication date source fielddata.attributes.publicationYear
Publication date reviewed2026-09-11
Primary categorysurgical_video
Resource rolecurrent_dataset
Dataset familycataract_lmm
Contained modalitiessurgical_video
Tasksclassification, segmentation, multilabel
Primary reported quantity3,000 videos
ClassesNot reported (Not reported)
Splitstrain, val, test
Size200.0 GB
Source-stated termsCC BY-NC-ND 4.0
Normalized termscc-by-nc-nd
Descriptive screening labelExplicit noncommercial clause recorded; check source
Terms scopedataset_files
Access frictionself_service_authenticated
Route backendHuggingFace Hub
Availabilityavailable (checked 2026-07-21)
Acquisition supportstandard_platform_supported
Legacy sample-loader statusMetadata and access only

Reported quantities

RoleCountUnitScopeBasisEvidence
Primary3,000videosPrimary quantity reported in the reviewed catalog sourcelegacy_catalog_fieldhuggingface.co/datasets

Counts retain their source-reported units. Additional rows can describe components, paired items, or derivative copies and are not automatically added to the primary quantity.

Notes

Largest public cataract-surgery-video resource by hours. CC BY-NC-ND 4.0 — non-commercial + no derivatives on HF.

Access information and download

# Read-only preflight
eyehub download cataract_lmm --data-dir ./data --dry-run --json

# Download, only when preflight reports supported behavior
eyehub download cataract_lmm --data-dir ./data

Upstream page: huggingface.co/datasets

Source-term evidence: huggingface.co/datasets

Loader status

This catalog record provides metadata and access instructions, but it does not yet include a standard DatasetSample loader. Inspect the source file structure or contribute a loader before using it in a training pipeline.

Citation

@misc{cataract_lmm,
title = { Cataract-LMM: Large-Scale Multi-Source Multi-Task Cataract Surgery Benchmark },
note = { Ahmadi MJ, Gandomi I, Abdi P, et al., 'Cataract-LMM: Large-scale multi-source multi-task benchmark for deep learning in surgical video analysis', Scientific Data 2026. doi:10.1038/s41597-026-07464-0 },
year = { 2026 },
url = { https://huggingface.co/datasets/mjahmadi/Cataract-LMM },
}

Source-stated terms

  • Raw source string: CC BY-NC-ND 4.0
  • Normalized category: cc-by-nc-nd
  • Apparent scope: dataset_files
  • Descriptive screening label: Explicit noncommercial clause recorded; check source

⚠️ Source-stated terms, scope, and normalized labels are curation metadata, not legal advice or a permission finding. Review the current official source before transfer or reuse.

Similar resources by shared modality

  • ophora: Ophora-160K: Ophthalmic Surgical Video Instruction Dataset (162,185 video clip instruction pairs, unknown)
  • lmod_plus: LMOD+ Multimodal Ophthalmology Benchmark (32,633 annotated instances, unknown)
  • insegcat: InSegCat: Instance Segmentation for Cataract Surgery (5,581 annotated images, research-only)
  • cadis: CaDIS: Cataract Dataset for Image Segmentation (4,670 images, research-only)
  • ophnet2024: OphNet2024: Ophthalmic Surgical Video Dataset (2,278 videos, cc-by-nc-sa)
  • lmod_cataract_1k: LMOD-Cataract-1K (2,256 images, cc-by)
  • lmod_cataract_1k_cot: Cataract-1K Surgical Analysis Chain-of-Thought Dataset (2,256 images, mit)
  • cataract1k: Cataract-1K: Large-Scale Cataract Surgery Video Dataset (1,000 videos, research-only)