---
title: "DINOv2: Learning Robust Visual Features without Supervision"
description: "DINOv2 trains one visual backbone to preserve both whole-image meaning and patch-level structure without captions or class labels. Follow LVD-142M curation, the moving teacher, masked patch targets, feature spreading, frozen-transfer evidence, and the data and bias boundaries around general-purpose features."
canonical_url: "https://fanout.sh/daily/2026-11-13-dinov2-visual-features"
md_url: "https://fanout.sh/daily/2026-11-13-dinov2-visual-features.md"
last_updated: "2026-11-13"
access: "public"
---

# DINOv2: Learning Robust Visual Features without Supervision

DINOv2 trains one visual backbone to preserve both whole-image meaning and patch-level structure without captions or class labels. Follow LVD-142M curation, the moving teacher, masked patch targets, feature spreading, frozen-transfer evidence, and the data and bias boundaries around general-purpose features.

## Paper details

- Authors: Maxime Oquab et al.

- Venue: TMLR 2024

- Track: AI Research

- Difficulty: Medium

- Reading time: 15 min

- Original paper: https://arxiv.org/abs/2304.07193v2

## What you will learn

- How visually retrieved web images turn a 1.2-billion-image pool into the curated LVD-142M pretraining set.

- Why crop-level self-distillation, masked patch prediction, and KoLeo feature spreading protect different uses of one representation.

- What frozen linear probes, dense prediction heads, distillation ablations, and the Dollar Street audit establish, and what private data still prevents reproducing.

## Continue exploring

[Compare image-text supervision](https://fanout.sh/daily/2026-08-23-clip-natural-language-supervision): Contrast DINOv2's image-only crop and patch targets with CLIP's alignment between an image and its natural-language caption.

---
This representation contains public Fanout content only. Protected Pro lessons, account data, billing, checkout, and pricing are not included.

Browse the public content map: https://fanout.sh/sitemap.md
