> ## Documentation Index
> Fetch the complete documentation index at: https://docs.generalrobotics.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# dinov2-base

> Extract DINOv2-base patch tokens and the [CLS] token for one RGB image

Extract DINOv2-base patch tokens and the \[CLS] token for one RGB image.

## Parameters

<ParamField body="image_input" type="ImageInput" required>
  RGB image as file path, URL, PIL Image, or numpy array.
</ParamField>

<ParamField body="timeout" type="float | None">
  Optional timeout in seconds for the HTTP request.
</ParamField>

## Returns

Dict\[str, np.ndarray]: Dict with `patch_tokens` of shape `(16, 16, 768)` and `cls` of shape
`(768,)`, both `float32` and L2-normalized per vector. Pool
`patch_tokens` over a foreground mask and re-normalize to compare crops
by cosine similarity without their background.

## Example

```python theme={null}
from grid_cortex_client import CortexClient, ModelType
import numpy as np
from PIL import Image
client = CortexClient()
crop = np.array(Image.open("lock_crop_224.png"))
mask = np.array(Image.open("lock_mask_224.png")) > 0      # (224, 224) bool
out = client.run(ModelType.DINOV2_BASE, image_input=crop)
out["patch_tokens"].shape, out["cls"].shape               # (16, 16, 768), (768,)
fg = mask.reshape(16, 14, 16, 14).mean((1, 3)) > 0.5      # (16, 16) patch mask
v = out["patch_tokens"][fg].mean(0)
v /= np.linalg.norm(v)                                    # cosine == v @ w
```


## Related topics

- [MDP Config](/simulation/isaac/session_configuration/mdp.md)
- [MobileManipulator](/python-api/robot-interface/mobilemanipulator.md)
- [Customizing RL Training](/simulation/isaac/reinforcement-learning/customizing-training.md)
- [Wheeled](/python-api/robot-interface/wheeled.md)
- [Arm](/python-api/robot-interface/arm.md)