Oquab and colleagues: DINOv2 self-supervised visual features

What changed

Scaled self-supervised vision representation learning with a curated image corpus and training recipe, reporting transferable visual features without task labels.

Scientific scope

Scaled self-supervised vision representation learning with a curated image corpus and training recipe, reporting transferable visual features without task labels. Scope limitation: Performance is evaluated on specified vision benchmarks; the project does not show that all downstream vision tasks require no supervision.

Evidence

The retained primary-source abstract, bibliographic record and stated review scope support this provisional assessment. No independent experimental replication by WRS is claimed. Performance is evaluated on specified vision benchmarks; the project does not show that all downstream vision tasks require no supervision.

Contributors

• The named authors listed in the linked research record · SCIENCE: Complete source byline retained in the research record; human research direction, method development and evaluation. Individual labor shares are not measured.

• Oquab and colleagues: DINOv2 self-supervised visual features — original paper
• DINOv2 official code and models — Meta FAIR

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction