README.md

---
pipeline_tag: image-segmentation
license: apple-amlr
base_model: apple/mobilevitv2-1.0-voc-deeplabv3
library_name: zeromodels
tags:
- keras
- zeromodels
- mobilevit
- deeplabv3
- image-segmentation
- semantic-segmentation
- arxiv:2110.02178
- arxiv:2206.02680
- pytorch
- jax
- tf
---

## ***See [our collection](https://huggingface.co/collections/zeromodels/mobilevit-v1-and-v2-6a8eaf6304112b66453f9ccc) for MobileViT / MobileViTV2 DeepLabV3 segmentation.***

# Run MobileViTV2 DeepLabV3 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-MobileViT%20DeepLabV3-blue)](https://imvision12.github.io/ZeroModels/mobilevitv2/) [![Collection](https://img.shields.io/badge/HF-MobileViT%20seg%20collection-yellow)](https://huggingface.co/collections/zeromodels/mobilevit-v1-and-v2-6a8eaf6304112b66453f9ccc)

# zeromodels/mobilevitv2_100_deeplabv3

Papers: [MobileViT (arXiv:{PAPER_ARXIV})]({PAPER_URL}) · [MobileViTV2 (arXiv:2206.02680)](https://arxiv.org/abs/2206.02680) · [HF Papers](https://huggingface.co/papers/2110.02178)

MobileViTV2 backbone + **DeepLabV3 ASPP** head for Pascal VOC semantic segmentation (21 classes). **Resolution is 512**, not the 256 used by ImageNet classification checkpoints. Always load the processor with `from_weights` so resize/crop match.

For more details on the model, please go to the upstream [model card](https://huggingface.co/apple/mobilevitv2-1.0-voc-deeplabv3).

Pure-**Keras 3** conversion of [`apple/mobilevitv2-1.0-voc-deeplabv3`](https://huggingface.co/apple/mobilevitv2-1.0-voc-deeplabv3) for [zeromodels](https://github.com/IMvision12/ZeroModels). One implementation runs unmodified on **TensorFlow / Torch / JAX**.

This is a **semantic segmentation** checkpoint (`MobileViTV2SemanticSegment`).

## ✨ Quick start

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

from PIL import Image
from zeromodels.models.mobilevitv2 import (
    MobileViTV2SemanticSegment,
    MobileViTV2ImageProcessor,
)

model = MobileViTV2SemanticSegment.from_weights("zeromodels/mobilevitv2_100_deeplabv3")
processor = MobileViTV2ImageProcessor.from_weights("zeromodels/mobilevitv2_100_deeplabv3")

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

Load any DeepLabV3 MobileViT variant the same way with `from_weights("zeromodels/<variant>")`:

| Variant | Hub | Family |
|---|---|---|
| `mobilevit_xxs_deeplabv3` | [`zeromodels/mobilevit_xxs_deeplabv3`](https://huggingface.co/zeromodels/mobilevit_xxs_deeplabv3) | MobileViT v1 |
| `mobilevit_xs_deeplabv3` | [`zeromodels/mobilevit_xs_deeplabv3`](https://huggingface.co/zeromodels/mobilevit_xs_deeplabv3) | MobileViT v1 |
| `mobilevit_s_deeplabv3` | [`zeromodels/mobilevit_s_deeplabv3`](https://huggingface.co/zeromodels/mobilevit_s_deeplabv3) | MobileViT v1 |
| `mobilevitv2_100_deeplabv3` | [`zeromodels/mobilevitv2_100_deeplabv3`](https://huggingface.co/zeromodels/mobilevitv2_100_deeplabv3) | MobileViT v2 |
| `mobilevitv2_150_deeplabv3` | [`zeromodels/mobilevitv2_150_deeplabv3`](https://huggingface.co/zeromodels/mobilevitv2_150_deeplabv3) | MobileViT v2 |

## Tips

- Set `KERAS_BACKEND` **before** importing Keras / zeromodels.
- Do not reuse a classification processor: seg checkpoints need 544/512.
- v1 imports from `mobilevit`; v2 from `mobilevitv2`.
- See [docs](https://imvision12.github.io/ZeroModels/mobilevitv2/) and [Loading Weights](https://imvision12.github.io/ZeroModels/loading_weights/).
- Upstream: `MobileViTV2SemanticSegment.from_weights("hf:apple/mobilevitv2-1.0-voc-deeplabv3")`.

## Special Thanks

A huge thank you to the Apple MobileViT authors for creating and releasing these models.

License: see YAML / upstream card.