---
pipeline_tag: image-classification
license: apache-2.0
base_model: timm/nextvit_small.bd_in1k
library_name: zeromodels
tags:
- keras
- zeromodels
- image-classification
- nextvit
- backbone
- arxiv:2207.05501
- pytorch
- jax
- tf
---

## ***See [our collection](https://huggingface.co/collections/zeromodels/next-vit-6a8eaece11e6326749bc87e4) for all versions of Next-ViT.***

# Run Next-ViT 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-Next--ViT-blue)](https://imvision12.github.io/ZeroModels/classification_backbones/) [![Collection](https://img.shields.io/badge/HF-Next--ViT%20collection-yellow)](https://huggingface.co/collections/zeromodels/next-vit-6a8eaece11e6326749bc87e4)

# zeromodels/nextvit_small_bd_in1k

Paper: [Next-ViT: Next Generation Vision Transformer for Efficient Deployment in Realistic Industrial Scenarios (arXiv:2207.05501)](https://arxiv.org/abs/2207.05501) · [HF Papers](https://huggingface.co/papers/2207.05501)

Next-ViT targets efficient industrial deployment with a hybrid CNN/Transformer stack. Classifier or hierarchical backbone.

For more details on the model, please go to the upstream [model card](https://huggingface.co/timm/nextvit_small.bd_in1k).

Pure-**Keras 3** conversion of [`timm/nextvit_small.bd_in1k`](https://huggingface.co/timm/nextvit_small.bd_in1k) for [zeromodels](https://github.com/IMvision12/ZeroModels). One implementation runs unmodified on **TensorFlow / Torch / JAX**.

This is an **image-classification / backbone** checkpoint (`NextViTImageClassify` / `NextViTModel`).

## ✨ Quick start

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

from PIL import Image
import numpy as np
from zeromodels.models.nextvit import NextViTImageClassify, NextViTModel

model = NextViTImageClassify.from_weights("zeromodels/nextvit_small_bd_in1k")
backbone = NextViTModel.from_weights(
    "zeromodels/nextvit_small_bd_in1k", as_backbone=True
)

image = Image.open("your_image.jpg").convert("RGB")
image = image.resize((224, 224))
x = np.asarray(image, dtype="float32")[None]  # (1, H, W, 3)
print(model(x).shape)  # (1, num_classes)
feats = backbone(x)
print(len(feats), [tuple(f.shape) for f in feats])
```

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

| Variant | Hub |
|---|---|
| `nextvit_base_bd_in1k` | [`zeromodels/nextvit_base_bd_in1k`](https://huggingface.co/zeromodels/nextvit_base_bd_in1k) |
| `nextvit_base_bd_in1k_384` | [`zeromodels/nextvit_base_bd_in1k_384`](https://huggingface.co/zeromodels/nextvit_base_bd_in1k_384) |
| `nextvit_base_bd_ssld_6m_in1k` | [`zeromodels/nextvit_base_bd_ssld_6m_in1k`](https://huggingface.co/zeromodels/nextvit_base_bd_ssld_6m_in1k) |
| `nextvit_base_bd_ssld_6m_in1k_384` | [`zeromodels/nextvit_base_bd_ssld_6m_in1k_384`](https://huggingface.co/zeromodels/nextvit_base_bd_ssld_6m_in1k_384) |
| `nextvit_large_bd_in1k` | [`zeromodels/nextvit_large_bd_in1k`](https://huggingface.co/zeromodels/nextvit_large_bd_in1k) |
| `nextvit_large_bd_in1k_384` | [`zeromodels/nextvit_large_bd_in1k_384`](https://huggingface.co/zeromodels/nextvit_large_bd_in1k_384) |
| `nextvit_large_bd_ssld_6m_in1k` | [`zeromodels/nextvit_large_bd_ssld_6m_in1k`](https://huggingface.co/zeromodels/nextvit_large_bd_ssld_6m_in1k) |
| `nextvit_large_bd_ssld_6m_in1k_384` | [`zeromodels/nextvit_large_bd_ssld_6m_in1k_384`](https://huggingface.co/zeromodels/nextvit_large_bd_ssld_6m_in1k_384) |
| `nextvit_small_bd_in1k` | [`zeromodels/nextvit_small_bd_in1k`](https://huggingface.co/zeromodels/nextvit_small_bd_in1k) |
| `nextvit_small_bd_in1k_384` | [`zeromodels/nextvit_small_bd_in1k_384`](https://huggingface.co/zeromodels/nextvit_small_bd_in1k_384) |
| `nextvit_small_bd_ssld_6m_in1k` | [`zeromodels/nextvit_small_bd_ssld_6m_in1k`](https://huggingface.co/zeromodels/nextvit_small_bd_ssld_6m_in1k) |
| `nextvit_small_bd_ssld_6m_in1k_384` | [`zeromodels/nextvit_small_bd_ssld_6m_in1k_384`](https://huggingface.co/zeromodels/nextvit_small_bd_ssld_6m_in1k_384) |

## Tips

- Set `KERAS_BACKEND` **before** importing Keras / zeromodels.
- `NextViTImageClassify` returns class logits; `NextViTModel` returns features (`as_backbone=True` for multi-scale stages).
- See [docs](https://imvision12.github.io/ZeroModels/classification_backbones/) and [Loading Weights](https://imvision12.github.io/ZeroModels/loading_weights/).
- Upstream / timm checkpoints: `NextViTImageClassify.from_weights("hf:timm/nextvit_small.bd_in1k")`.

## Special Thanks

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

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