README.md

---
pipeline_tag: mask-generation
license: other
license_name: sam-license
license_link: https://github.com/facebookresearch/sam3/blob/main/LICENSE
base_model: facebook/sam3
library_name: zeromodels
tags:
- keras
- zeromodels
- sam3
- mask-generation
- image-segmentation
- arxiv:2511.16719
- pytorch
- jax
- tf
---

## ***See [our collection](https://huggingface.co/collections/zeromodels/sam-v1-v2-v3-6a8eaf73a274fffa2ea83cfd) for all versions of SAM.***

# Run SAM3 with Keras 3: JAX, PyTorch, or TensorFlow

[![GitHub](https://img.shields.io/badge/GitHub-ZeroModels-black?logo=github)](https://github.com/IMvision12/ZeroModels) [![Docs](https://img.shields.io/badge/Docs-SAM3-blue)](https://imvision12.github.io/ZeroModels/sam3/) [![Collection](https://img.shields.io/badge/HF-SAM%20collection-yellow)](https://huggingface.co/collections/zeromodels/sam-v1-v2-v3-6a8eaf73a274fffa2ea83cfd)

# zeromodels/sam3

Paper: [SAM 3: Segment Anything with Concepts (arXiv:2511.16719)](https://arxiv.org/abs/2511.16719) · [HF Papers](https://huggingface.co/papers/2511.16719)

SAM3 segments by concept, not location: give it a noun phrase and it finds every matching instance. A ViT-L backbone and FPN feed a DETR-style encoder/decoder with object queries; a CLIP text encoder supplies the open-vocabulary side. Boxes can still be mixed with text.

For more details on the model, please go to Meta's original [model card](https://huggingface.co/facebook/sam3).

Pure-**Keras 3** conversion of [`facebook/sam3`](https://huggingface.co/facebook/sam3) for [zeromodels](https://github.com/IMvision12/ZeroModels). One implementation runs unmodified on **TensorFlow / Torch / JAX**.

This is a **concept-prompted** checkpoint (`SAM3InstanceSegment` / `SAM3Detect` / `SAM3SemanticSegment`): pass a text noun phrase (backbone ViT-L/14).

## ✨ Quick start

```python
import os
os.environ["KERAS_BACKEND"] = "torch"  # or "jax" / "tensorflow"

from zeromodels.models.sam3 import SAM3InstanceSegment

segmenter = SAM3InstanceSegment.from_weights("zeromodels/sam3")
result = segmenter.predict(
    images="your_image.jpg", text="person", threshold=0.3
)[0]
print(len(result["scores"]), result["masks"].shape)
```

Load any SAM / SAM2 / SAM3 variant the same way with `from_weights("zeromodels/<variant>")` (use `SAM3InstanceSegment` for this repo):

| Variant | Hub | Family |
|---|---|---|
| `sam_vit_base` | [`zeromodels/sam_vit_base`](https://huggingface.co/zeromodels/sam_vit_base) | SAM |
| `sam_vit_large` | [`zeromodels/sam_vit_large`](https://huggingface.co/zeromodels/sam_vit_large) | SAM |
| `sam_vit_huge` | [`zeromodels/sam_vit_huge`](https://huggingface.co/zeromodels/sam_vit_huge) | SAM |
| `sam2_hiera_small` | [`zeromodels/sam2_hiera_small`](https://huggingface.co/zeromodels/sam2_hiera_small) | SAM2 |
| `sam2_hiera_base_plus` | [`zeromodels/sam2_hiera_base_plus`](https://huggingface.co/zeromodels/sam2_hiera_base_plus) | SAM2 |
| `sam2_hiera_large` | [`zeromodels/sam2_hiera_large`](https://huggingface.co/zeromodels/sam2_hiera_large) | SAM2 |
| `sam3` | [`zeromodels/sam3`](https://huggingface.co/zeromodels/sam3) | SAM3 |

## Tips

- Set `KERAS_BACKEND` **before** importing Keras / zeromodels.
- SAM / SAM2: point coordinates are in original pixel space; box prompts need `enable_boxes=True` / `include_box_input=True` when building the graph.
- SAM2 in this port is image-only (no video memory bank).
- SAM3: prefer `SAM3InstanceSegment.predict(...)` for text prompts; upstream `facebook/sam3` is gated.
- See [SAM3 docs](https://imvision12.github.io/ZeroModels/sam3/) and [Loading Weights](https://imvision12.github.io/ZeroModels/loading_weights/).
- Community / upstream safetensors still work via the `hf:` prefix, e.g. `SAM3Model.from_weights("hf:facebook/sam3")`.

## Special Thanks

A huge thank you to the Meta SAM 3 authors for creating and releasing these models.

License: see the [SAM 3 LICENSE](https://github.com/facebookresearch/sam3/blob/main/LICENSE) (Hub tag: `other` / `sam-license`). Upstream [`facebook/sam3`](https://huggingface.co/facebook/sam3) is gated.