dinov3-vith16plus-torch

DINOv3 ViT-H+/16 self-supervised image embeddings pretrained on LVD-1689M (1.7B images). Requires accepting the DINOv3 license on Hugging Face.

Details

• Model name: dinov3-vith16plus-torch

• Model source: https://huggingface.co/facebook/dinov3-vith16plus-pretrain-lvd1689m

• Model author: Oriane Simeoni, et al.

• Model license: DINOv3 License

• Model size: 3.13 GB

• Exposes embeddings? yes

• Tags: embeddings, torch, dinov3, transformer, official

Requirements

• Packages: torch, torchvision, transformers>=4.56

• CPU support

  • yes

• GPU support

  • yes

Example usage

import fiftyone as fo
import fiftyone.zoo as foz

dataset = foz.load_zoo_dataset(
    "coco-2017",
    split="validation",
    dataset_name=fo.get_default_dataset_name(),
    max_samples=50,
    shuffle=True,
)

model = foz.load_zoo_model("dinov3-vith16plus-torch")

dataset.apply_model(model, label_field="predictions")

session = fo.launch_app(dataset)