DINOv2

lastmod 2026-01-19created 2023-05-266 edits

PyTorch code and models for the DINOv2 self-supervised learning method.

A computer vision model released by Meta AI that is capable of self-supervised learning

Features

• A new method for training high-performance computer vision models
• Delivers strong performance without requiring fine-tuning, so it can be used as a backbone for various computer vision tasks
  • Using Self-Supervision, it can learn from any image collection, and can also learn new features such as depth estimation that are not possible with current approaches
• PyTorch code and models released as open source

See also

• Deep learning
• Meta
• Self-Supervised Vision Transformers with DINO (DiNO)
• Exponential moving average (EMA)
• RF-DETR (Object Detection)

Favorite site

• DINOv2: State-of-the-art computer vision models with self-supervised learning
• facebookresearch/dinov2
• (Paper review) DINOv2: Learning Robust Visual Features without Supervision explained
• (Paper review) DINOv2: Learning Robust Visual Features without Supervision - That Time I Got Reincarnated as an AI