---
pipeline_tag: image-classification
license: other
base_model: timm/mobilevitv2_050.cvnets_in1k
library_name: zeromodels
tags:
- keras
- zeromodels
- image-classification
- mobilevitv2
- backbone
- arxiv:2206.02680
- pytorch
- jax
- tf
---

## ***See [our collection](https://huggingface.co/collections/zeromodels/mobilevit-v2-6a8eae53d6249f084e6875fe) for all versions of MobileViT-V2.***

# Run MobileViT-V2 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--V2-blue)](https://imvision12.github.io/ZeroModels/mobilevitv2/) [![Collection](https://img.shields.io/badge/HF-MobileViT--V2%20collection-yellow)](https://huggingface.co/collections/zeromodels/mobilevit-v2-6a8eae53d6249f084e6875fe)

# zeromodels/mobilevitv2_050_cvnets_in1k

Paper: [Separable Self-attention for Mobile Vision Transformers (arXiv:2206.02680)](https://arxiv.org/abs/2206.02680) · [HF Papers](https://huggingface.co/papers/2206.02680)

MobileViTV2 replaces MHSA with separable self-attention (O(k)) and scales width via a single multiplier. Classification at 256/384; DeepLabV3 segmentation is a separate script/collection.

For more details on the model, please go to the upstream [model card](https://huggingface.co/timm/mobilevitv2_050.cvnets_in1k).

Pure-**Keras 3** conversion of [`timm/mobilevitv2_050.cvnets_in1k`](https://huggingface.co/timm/mobilevitv2_050.cvnets_in1k) for [zeromodels](https://github.com/IMvision12/ZeroModels). One implementation runs unmodified on **TensorFlow / Torch / JAX**.

This is an **image-classification / backbone** checkpoint (`MobileViTV2ImageClassify` / `MobileViTV2Model`).

## ✨ Quick start

```python
import os

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

from PIL import Image
from zeromodels.models.mobilevitv2 import MobileViTV2ImageClassify, MobileViTV2Model, MobileViTV2ImageProcessor

model = MobileViTV2ImageClassify.from_weights("zeromodels/mobilevitv2_050_cvnets_in1k")
processor = MobileViTV2ImageProcessor.from_weights("zeromodels/mobilevitv2_050_cvnets_in1k")

image = Image.open("your_image.jpg").convert("RGB")
pixels = processor(image)  # resize + normalize (normalization lives in the processor)
logits = model(pixels, training=False)
print(logits.shape)  # (1, num_classes)

# Feature extraction: the backbone without the classifier head
backbone = MobileViTV2Model.from_weights("zeromodels/mobilevitv2_050_cvnets_in1k", as_backbone=True)
features = backbone(pixels, training=False)
```

Load any MobileViT-V2 variant the same way with `from_weights("zeromodels/<variant>")`:

| Variant | Hub |
|---|---|
| `mobilevitv2_050_cvnets_in1k` | [`zeromodels/mobilevitv2_050_cvnets_in1k`](https://huggingface.co/zeromodels/mobilevitv2_050_cvnets_in1k) |
| `mobilevitv2_075_cvnets_in1k` | [`zeromodels/mobilevitv2_075_cvnets_in1k`](https://huggingface.co/zeromodels/mobilevitv2_075_cvnets_in1k) |
| `mobilevitv2_100_cvnets_in1k` | [`zeromodels/mobilevitv2_100_cvnets_in1k`](https://huggingface.co/zeromodels/mobilevitv2_100_cvnets_in1k) |
| `mobilevitv2_125_cvnets_in1k` | [`zeromodels/mobilevitv2_125_cvnets_in1k`](https://huggingface.co/zeromodels/mobilevitv2_125_cvnets_in1k) |
| `mobilevitv2_150_cvnets_in1k` | [`zeromodels/mobilevitv2_150_cvnets_in1k`](https://huggingface.co/zeromodels/mobilevitv2_150_cvnets_in1k) |
| `mobilevitv2_150_cvnets_in22k_ft_in1k` | [`zeromodels/mobilevitv2_150_cvnets_in22k_ft_in1k`](https://huggingface.co/zeromodels/mobilevitv2_150_cvnets_in22k_ft_in1k) |
| `mobilevitv2_150_cvnets_in22k_ft_in1k_384` | [`zeromodels/mobilevitv2_150_cvnets_in22k_ft_in1k_384`](https://huggingface.co/zeromodels/mobilevitv2_150_cvnets_in22k_ft_in1k_384) |
| `mobilevitv2_175_cvnets_in1k` | [`zeromodels/mobilevitv2_175_cvnets_in1k`](https://huggingface.co/zeromodels/mobilevitv2_175_cvnets_in1k) |
| `mobilevitv2_175_cvnets_in22k_ft_in1k` | [`zeromodels/mobilevitv2_175_cvnets_in22k_ft_in1k`](https://huggingface.co/zeromodels/mobilevitv2_175_cvnets_in22k_ft_in1k) |
| `mobilevitv2_175_cvnets_in22k_ft_in1k_384` | [`zeromodels/mobilevitv2_175_cvnets_in22k_ft_in1k_384`](https://huggingface.co/zeromodels/mobilevitv2_175_cvnets_in22k_ft_in1k_384) |
| `mobilevitv2_200_cvnets_in1k` | [`zeromodels/mobilevitv2_200_cvnets_in1k`](https://huggingface.co/zeromodels/mobilevitv2_200_cvnets_in1k) |
| `mobilevitv2_200_cvnets_in22k_ft_in1k` | [`zeromodels/mobilevitv2_200_cvnets_in22k_ft_in1k`](https://huggingface.co/zeromodels/mobilevitv2_200_cvnets_in22k_ft_in1k) |
| `mobilevitv2_200_cvnets_in22k_ft_in1k_384` | [`zeromodels/mobilevitv2_200_cvnets_in22k_ft_in1k_384`](https://huggingface.co/zeromodels/mobilevitv2_200_cvnets_in22k_ft_in1k_384) |

## Tips

- Set `KERAS_BACKEND` **before** importing Keras / zeromodels.
- `MobileViTV2ImageClassify` returns class logits; `MobileViTV2Model` returns features (`as_backbone=True` for multi-scale stages).
- See [docs](https://imvision12.github.io/ZeroModels/mobilevitv2/) and [Loading Weights](https://imvision12.github.io/ZeroModels/loading_weights/).
- Upstream / timm checkpoints: `MobileViTV2ImageClassify.from_weights("hf:timm/mobilevitv2_050.cvnets_in1k")`.

## Special Thanks

A huge thank you to the MobileViT-V2 authors and the timm / Hub communities for creating and releasing these models.

License: see YAML `license` (usually matches the upstream checkpoint).