---
pipeline_tag: image-segmentation
license: mit
base_model: facebook/mask2former-swin-base-coco-instance
library_name: kerasformers
tags:
- keras
- kerasformers
- mask2former
- instance-segmentation
- image-segmentation
- arxiv:2112.01527
- pytorch
- jax
- tf
---

## ***See [our collection](https://huggingface.co/collections/kerasformers/mask2former-6a6a8f4248526cc9a74b9b97) for all versions of Mask2Former.***

# Run Mask2Former with Keras 3: JAX, PyTorch, or TensorFlow

[![GitHub](https://img.shields.io/badge/GitHub-KerasFormers-black?logo=github)](https://github.com/IMvision12/KerasFormers) [![Docs](https://img.shields.io/badge/Docs-Mask2Former-blue)](https://imvision12.github.io/KerasFormers/mask2former/) [![Collection](https://img.shields.io/badge/HF-Mask2Former%20collection-yellow)](https://huggingface.co/collections/kerasformers/mask2former-6a6a8f4248526cc9a74b9b97)

# kerasformers/mask2former-swin-base-coco-instance

Paper: [Masked-attention Mask Transformer for Universal Image Segmentation (arXiv:2112.01527)](https://arxiv.org/abs/2112.01527) · [HF Papers](https://huggingface.co/papers/2112.01527)

Mask2Former improves MaskFormer with masked attention in the transformer decoder, restricting cross-attention to predicted mask regions for sharper boundaries and stronger universal segmentation.

For more details on the model, please go to the upstream [model card](https://huggingface.co/facebook/mask2former-swin-base-coco-instance).

Pure-**Keras 3** conversion of [`facebook/mask2former-swin-base-coco-instance`](https://huggingface.co/facebook/mask2former-swin-base-coco-instance) for [kerasformers](https://github.com/IMvision12/KerasFormers). One implementation runs unmodified on **TensorFlow / Torch / JAX**.

This is a **instance** checkpoint (`Mask2FormerUniversalSegment`) (trained for instance; architecture is universal).

## ✨ Quick start

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

from PIL import Image
from kerasformers.models.mask2former import Mask2FormerUniversalSegment, Mask2FormerImageProcessor

model = Mask2FormerUniversalSegment.from_weights("kerasformers/mask2former-swin-base-coco-instance")
processor = Mask2FormerImageProcessor.from_weights("kerasformers/mask2former-swin-base-coco-instance")

image = Image.open("your_image.jpg").convert("RGB")
output = model(processor(image)["pixel_values"], training=False)
result = processor.post_process_panoptic_segmentation(
    output, target_size=(image.height, image.width)
)
print(result["segmentation"].shape)
```

Load any Mask2Former variant the same way with `from_weights("kerasformers/<variant>")`:

| Variant | Hub | Task |
|---|---|---|
| `mask2former-swin-tiny-coco-instance` | [`kerasformers/mask2former-swin-tiny-coco-instance`](https://huggingface.co/kerasformers/mask2former-swin-tiny-coco-instance) | instance |
| `mask2former-swin-small-coco-instance` | [`kerasformers/mask2former-swin-small-coco-instance`](https://huggingface.co/kerasformers/mask2former-swin-small-coco-instance) | instance |
| `mask2former-swin-base-coco-instance` | [`kerasformers/mask2former-swin-base-coco-instance`](https://huggingface.co/kerasformers/mask2former-swin-base-coco-instance) | instance |
| `mask2former-swin-large-coco-instance` | [`kerasformers/mask2former-swin-large-coco-instance`](https://huggingface.co/kerasformers/mask2former-swin-large-coco-instance) | instance |
| `mask2former-swin-tiny-coco-panoptic` | [`kerasformers/mask2former-swin-tiny-coco-panoptic`](https://huggingface.co/kerasformers/mask2former-swin-tiny-coco-panoptic) | panoptic |
| `mask2former-swin-tiny-ade-semantic` | [`kerasformers/mask2former-swin-tiny-ade-semantic`](https://huggingface.co/kerasformers/mask2former-swin-tiny-ade-semantic) | semantic |

## Tips

- Set `KERAS_BACKEND` **before** importing Keras / kerasformers.
- The task suffix is what the checkpoint was trained for; post-process accordingly.
- See [Mask2Former docs]({DOCS_URL}) and [Loading Weights](https://imvision12.github.io/KerasFormers/loading_weights/).
- Community / upstream weights: `Mask2FormerUniversalSegment.from_weights("hf:facebook/mask2former-swin-base-coco-instance")`.

## Special Thanks

A huge thank you to the Facebook AI Research Mask2Former authors for creating and releasing these models.

License: MIT.