Model type
Vision transformer; this record is the paper-specific evaluated configuration.
Configuration
Cell-DINO ViT-L
Cell-DINO ViT-L learns representations of fluorescent cell images, including protein-localisation information.
Sources
Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · Results/Cell-DINO outperforms standard supervised ViT (paragraph 3); Results/Cell-DINO outperforms alternative self-supervised strategies (paragraph 3)How it worksEvaluated procedure (conceptual)
Conceptual input–method–output guide. Check the procedure text and linked evaluation for fitted components, additional inputs and exact settings.
Sources
Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · Results/Cell-DINO is competitive against highly tuned models (paragraph 1); Results/Cell-DINO outperforms standard supervised ViT (paragraph 3)At a glance
Biological inputs
Fluorescence-microscopy cell images
Sources
Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · Methods/Cell-DINO algorithm (paragraph 1); Methods/Datasets (paragraph 5)Outputs
Cell-image embeddings and downstream protein-localisation predictions
Sources
Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · Results/Cell-DINO outperforms standard supervised ViT (paragraph 3); Results/Cell-DINO outperforms standard supervised ViT (paragraph 1)limited source coverage · Automated source review, 2026-09-16. All specifications and missing details
Evaluations and results
Release 2026-09-17-d277315f7d76 · 1 evaluation · 1 metric row. Different protocols are not a single leaderboard.
| Metric and finding | Coverage and uncertainty | Evidence |
|---|---|---|
| Cell-DINO ViT-L: protein localization classification Self-supervised microscopy embedding pre-trained on HPA-FoV; downstream protein-localization classifier. Dataset-specific pretraining; the paper does not claim a general-purpose foundation model that generalizes beyond these benchmarks. Author-reported evaluation · Evaluation metadata: needs review | ||
| 65.5% F1 Unit: percent · Direction: unknown | Uncertainty: not reported in legacy extract Scored: Not reported · Eligible: Not reported | source checkedCell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · Table 2, HPA-FoV section, Cell-DINO row, PL column Source checking is not independent reproduction. |
How it works
How the evaluated method works
DINOv2 self-supervised learning trains a vision transformer on cell images. Frozen embeddings support downstream classifiers; the protein-localisation experiment includes a separately trained two-layer classifier.
Sources
Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · Results/Cell-DINO is competitive against highly tuned models (paragraph 1); Results/Cell-DINO outperforms standard supervised ViT (paragraph 3)What was evaluated
The linked evaluation record identifies Cell-DINO ViT-L: protein localization classification. Its dataset, split, adaptation and evidence origin remain attached to the reported results.
Sources
Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · The named evaluation’s methods and comparison table; exact preserved evaluation IDs: evaluation-lit-b4-015Strengths and limitations
Strengths and considerations
- Self-supervised representations can support protein-localisation classification when annotations are limited.
Sources
Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · Abstract (paragraph 1); Methods/Supervised ViTs (paragraph 1)
Limitations and conditions
- This record is included for molecular protein-localisation evaluation; performance on these microscopy datasets is not evidence for general clinical imaging.
Sources
Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · Results/Cell-DINO outperforms other pre-trained models (paragraph 3); Results/Cell-DINO outperforms alternative self-supervised strategies (paragraph 3)
Profile review details
Primary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction.
Stable record: reported-model-808b23c65fbc89Specifications
Inputs, training, access and other details
Explanatory profile: limited source coverage · Automated source review, 2026-09-16. Review applies to the cited claims; unresolved fields are listed below. Numerical results retain their own review status.
| Property | Description and evidence |
|---|---|
| Model type | Vision transformer; this record is the paper-specific evaluated configuration.SourcesCell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · Results/Cell-DINO is competitive against highly tuned models (paragraph 1); Results/Cell-DINO outperforms standard supervised ViT (paragraph 3) |
| Architecture / procedure | DINOv2 self-supervised learning trains a vision transformer on cell images. Frozen embeddings support downstream classifiers; the protein-localisation experiment includes a separately trained two-layer classifier.SourcesCell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · Results/Cell-DINO is competitive against highly tuned models (paragraph 1); Results/Cell-DINO outperforms standard supervised ViT (paragraph 3) |
| Biological inputs | Fluorescence-microscopy cell imagesSourcesCell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · Methods/Cell-DINO algorithm (paragraph 1); Methods/Datasets (paragraph 5) |
| Outputs | Cell-image embeddings and downstream protein-localisation predictionsSourcesCell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · Results/Cell-DINO outperforms standard supervised ViT (paragraph 3); Results/Cell-DINO outperforms standard supervised ViT (paragraph 1) |
| Parameters | The HPA-SC downstream two-layer classifier has 1.2 million supervised parameters; this is not the total of the frozen ViT-L encoder plus classifier.SourcesCell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · Results/Cell-DINO reduces the dependency on manual annotations (paragraph 1); Results/Cell-DINO is competitive against highly tuned models (paragraph 1) |
| Known versions / configuration | Cell-DINO ViT-L is the comparison-table label; that label does not specify an immutable weight revision. · Not reported in inspected sourcesSourcesCell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · Model identification in the comparison table and corresponding Methods; immutable checkpoint revision is not supplied by the table label. |
| Training data / fitting | Separate ViT-L pretraining datasets: HPA-FoV about 200,000 images, HPA-SC about 500,000 images, and combined Cell Painting about five million images. The protein-localisation classifier is trained separately.SourcesCell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · Methods/Computational resources used for training Cell-DINO models (paragraph 1); Methods/Datasets (paragraph 4) |
| Context limits | A maximum input/context length for this exact evaluated configuration is not established by the inspected sources. · Not reported in inspected sourcesSources (2)Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy; facebookresearch/dinov2 docs/README_CELL_DINO.md · Methods/Datasets; Methods/Cell-DINO algorithm; Methods/Vision Transformer (ViT) architecture; Methods/Supervised ViTs; Methods/Training and evaluation protocol on HPA datasets; Methods/Kaggle evaluation protocol; Methods/Cell Painting evaluation protocol; Methods/Baselines; inspected for explicit maximum input length (dataset lengths and family-wide limits are not substituted); docs/README_CELL_DINO.md at pinned repository revision |
| Access | Official study implementation and usage documentation: https://github.com/facebookresearch/dinov2/blob/7764ea0f912e53c92e82eb78a2a1631e92725fc8/docs/README_CELL_DINO.md. This pinned documentation revision is not automatically the evaluated weight revision.Sourcesfacebookresearch/dinov2 docs/README_CELL_DINO.md · docs/README_CELL_DINO.md; installation, model download and usage instructions |
| Code licence | Conflicting official statements: the Cell-DINO README says CC BY-NC, but its linked code licence is headed Creative Commons Attribution 4.0 International. Code reuse terms require author clarification; the DINOv2 root licence is not substituted. · Not reported in inspected sourcesSources (2)facebookresearch/dinov2 docs/README_CELL_DINO.md; facebookresearch/dinov2 LICENSE_CELL_DINO_CODE · docs/README_CELL_DINO.md / License; linked LICENSE_CELL_DINO_CODE heading and grant |
| Weights licence | FAIR Noncommercial Research License, version 1 (18 August 2025), for the Cell-DINO model materials.Sourcesfacebookresearch/dinov2 LICENSE_CELL_DINO_MODELS · LICENSE_CELL_DINO_MODELS; title, definitions and Section 1 |
Evidence table
Inspect claims, sources and review details
Trace each statement to its source and review. A context-only reference supports the record generally; it does not verify an individual field. Source checking does not reproduce an experiment.
One row per statement and cited source. Multiple citations are not independent evaluations. Shared locators are labelled explicitly.
21 evidence rows matching the loaded filters
| Property and statement | Original source and location | Review and provenance |
|---|---|---|
| Diagram caption Conceptual input–method–output guide. Check the procedure text and linked evaluation for fitted components, additional inputs and exact settings. Individual claims | Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy Results/Cell-DINO is competitive against highly tuned models (paragraph 1); Results/Cell-DINO outperforms standard supervised ViT (paragraph 3) Version: version of record | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Diagram steps ["Fluorescence-microscopy cell images","Cell-DINO ViT-L","Cell-image embeddings and downstream protein-localisation predictions"] Individual claims | Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy Results/Cell-DINO is competitive against highly tuned models (paragraph 1); Results/Cell-DINO outperforms standard supervised ViT (paragraph 3) Version: version of record | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Diagram title Evaluated procedure (conceptual) Individual claims | Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy Results/Cell-DINO is competitive against highly tuned models (paragraph 1); Results/Cell-DINO outperforms standard supervised ViT (paragraph 3) Version: version of record | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Model type Vision transformer; this record is the paper-specific evaluated configuration. Individual claims | Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy Results/Cell-DINO is competitive against highly tuned models (paragraph 1); Results/Cell-DINO outperforms standard supervised ViT (paragraph 3) Version: version of record | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Architecture / procedure DINOv2 self-supervised learning trains a vision transformer on cell images. Frozen embeddings support downstream classifiers; the protein-localisation experiment includes a separately trained two-layer classifier. Individual claims | Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy Results/Cell-DINO is competitive against highly tuned models (paragraph 1); Results/Cell-DINO outperforms standard supervised ViT (paragraph 3) Version: version of record | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Weights licence FAIR Noncommercial Research License, version 1 (18 August 2025), for the Cell-DINO model materials. Individual claims | facebookresearch/dinov2 LICENSE_CELL_DINO_MODELS LICENSE_CELL_DINO_MODELS; title, definitions and Section 1 Version: 7764ea0f912e53c92e82eb78a2a1631e92725fc8 | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Biological inputs Fluorescence-microscopy cell images Individual claims | Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy Methods/Cell-DINO algorithm (paragraph 1); Methods/Datasets (paragraph 5) Version: version of record | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Outputs Cell-image embeddings and downstream protein-localisation predictions Individual claims | Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy Results/Cell-DINO outperforms standard supervised ViT (paragraph 3); Results/Cell-DINO outperforms standard supervised ViT (paragraph 1) Version: version of record | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Parameters The HPA-SC downstream two-layer classifier has 1.2 million supervised parameters; this is not the total of the frozen ViT-L encoder plus classifier. Individual claims | Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy Results/Cell-DINO reduces the dependency on manual annotations (paragraph 1); Results/Cell-DINO is competitive against highly tuned models (paragraph 1) Version: version of record | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Known versions / configuration Cell-DINO ViT-L is the comparison-table label; that label does not specify an immutable weight revision. Individual claims | Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy Model identification in the comparison table and corresponding Methods; immutable checkpoint revision is not supplied by the table label. Version: version of record | unreported automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
Sources and history
Release 2026-09-17-d277315f7d76 · Record review: needs review
4 source records and release history
- Cell-DINO: Self-supervised image-based embeddings for cell fluorescent microscopy · Original source · version of record
- facebookresearch/dinov2 docs/README_CELL_DINO.md · Original source · 7764ea0f912e53c92e82eb78a2a1631e92725fc8
- facebookresearch/dinov2 LICENSE_CELL_DINO_CODE · Original source · 7764ea0f912e53c92e82eb78a2a1631e92725fc8
- facebookresearch/dinov2 LICENSE_CELL_DINO_MODELS · Original source · 7764ea0f912e53c92e82eb78a2a1631e92725fc8
Technical metadata and extraction receipts
Stable ID: reported-model-808b23c65fbc89
- areas
- cells-tissues
- entity level
- method
- version
- Not reported
- reported name
- Cell-DINO ViT-L
- historical missing metadata
- version: not_reported_in_legacy_extract; checkpoint revision: not_reported_in_legacy_extract; training data: not_reported_in_legacy_extract; licence: not_reported_in_legacy_extract
- metadata review scope
- historical_missing_metadata preserves the original discovery state. Current descriptive evidence and missingness are recorded in profile.facts; numerical-result review is separate.
- legacy kinds
- model
- entity classification
- review date: 2026-09-17; rationale: This source-scoped entry preserves the method/configuration actually named in an evaluation. It is neither a global family identity nor proof of an immutable checkpoint; the linked evaluation retains adaptation, fitting and scoring details.; source ids: cell-dino-2025; source locator: Results/Cell-DINO is competitive against highly tuned models (paragraph 1); Results/Cell-DINO outperforms standard supervised ViT (paragraph 3) | Results/Cell-DINO outperforms standard supervised ViT (paragraph 3); Results/Cell-DINO outperforms alternative self-supervised strategies (paragraph 3); ambiguities: Configuration means the source-labelled evaluated identity. It does not establish missing checkpoint hashes, default settings or equivalence to same-named records in other papers.