---
pipeline_tag: image-classification
license: apache-2.0
base_model: Zetatech/pvt-medium-224
library_name: zeromodels
tags:
- keras
- zeromodels
- image-classification
- pvt
- backbone
- arxiv:2102.12122
- pytorch
- jax
- tf
---

## ***See [our collection](https://huggingface.co/collections/zeromodels/pvt-and-pvtv2-6a90e9dd0a2b03a982d0b876) for all PVT and PVTv2 versions.***

# Run PVT 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-PVT-blue)](https://imvision12.github.io/ZeroModels/pvt/) [![Collection](https://img.shields.io/badge/HF-PVT%20collection-yellow)](https://huggingface.co/collections/zeromodels/pvt-and-pvtv2-6a90e9dd0a2b03a982d0b876)

# zeromodels/pvt-medium-224

Paper: [Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions (arXiv:2102.12122)](https://arxiv.org/abs/2102.12122) · [HF Papers](https://huggingface.co/papers/2102.12122)

PVT is a hierarchical vision transformer: four pyramid stages with spatial-reduction attention over non-overlapping patches and learned position embeddings. Use `PvtImageClassify` for logits or `PvtModel` for tokens / per-stage features via `as_backbone=True`.

- Parameters: ~44.2M
- ImageNet-1k top-1: **81.2%**

For more details on the model, see the upstream [model card](https://huggingface.co/Zetatech/pvt-medium-224).

Pure-**Keras 3** conversion of [`Zetatech/pvt-medium-224`](https://huggingface.co/Zetatech/pvt-medium-224) for [zeromodels](https://github.com/IMvision12/ZeroModels). One implementation runs unmodified on **TensorFlow / Torch / JAX**.

This is an **image-classification / backbone** checkpoint (`PvtImageClassify` / `PvtModel`).

## ✨ Quick start

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

from PIL import Image
import numpy as np
from zeromodels.models.pvt import PvtImageClassify, PvtModel

model = PvtImageClassify.from_weights("zeromodels/pvt-medium-224")
backbone = PvtModel.from_weights("zeromodels/pvt-medium-224", as_backbone=True)

image = Image.open("your_image.jpg").convert("RGB").resize((224, 224))
x = np.asarray(image, dtype="float32")[None]  # (1, H, W, 3), raw [0, 255]
print(model(x).shape)  # (1, num_classes)
feats = backbone(x)
print(len(feats), [tuple(f.shape) for f in feats])  # 4-stage feature pyramid
```

Normalization is baked into the graph, so pass raw `[0, 255]` pixels. Load any PVT
variant the same way with `from_weights("zeromodels/<variant>")`:

| Variant | ImageNet-1k top-1 | Hub |
|---|---|---|
| `pvt-tiny-224` | 75.1% | [`zeromodels/pvt-tiny-224`](https://huggingface.co/zeromodels/pvt-tiny-224) |
| `pvt-small-224` | 79.8% | [`zeromodels/pvt-small-224`](https://huggingface.co/zeromodels/pvt-small-224) |
| `pvt-medium-224` | 81.2% | [`zeromodels/pvt-medium-224`](https://huggingface.co/zeromodels/pvt-medium-224) |
| `pvt-large-224` | 81.7% | [`zeromodels/pvt-large-224`](https://huggingface.co/zeromodels/pvt-large-224) |

## Tips

- Set `KERAS_BACKEND` **before** importing Keras / zeromodels.
- `PvtImageClassify` returns class logits; `PvtModel` returns features (`as_backbone=True` for the four-stage pyramid).
- Both the model and its data format (`channels_last` / `channels_first`) are supported and bit-exact.
- See the [docs](https://imvision12.github.io/ZeroModels/pvt/) and [Loading Weights](https://imvision12.github.io/ZeroModels/loading_weights/).
- Upstream checkpoints load directly: `PvtImageClassify.from_weights("hf:Zetatech/pvt-medium-224")`.

## Special Thanks

A huge thank you to the PVT authors ([whai362/PVT](https://github.com/whai362/PVT)) and the Hugging Face
community for creating and releasing these models.

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