---
title: sapiens2-pretrain-1b-4k
canonical_url: "https://www.modelscope.cn/models/facebook/sapiens2-pretrain-1b-4k"
md_url: "https://www.modelscope.cn/models/facebook/sapiens2-pretrain-1b-4k.md"
repository: facebook/sapiens2-pretrain-1b-4k
last_updated: 2026-10-01
license: other
library_name:
  - pytorch
  - safetensors
frameworks:
  - pytorch
downloads: 10
stars: 0
tags:
  - sapiens
  - sapiens2
  - vision-transformer
  - human-centric
  - pretrained-backbone
  - feature-extraction
  - 4k
---

# sapiens2-pretrain-1b-4k

> sapiens2-pretrain-1b-4k - facebook 在 ModelScope 开源的模型。Sapiens2 is a family of high-resolution vision transformers pretrained on 1 billion human images — designed for human-centric tasks such as pose estimation, body-part segmentation, surface normals, and…

facebook/sapiens2-pretrain-1b-4k 是 ModelScope 魔搭社区上的机器学习模型，采用 other 许可。

- **Repository**: facebook/sapiens2-pretrain-1b-4k
- **License**: other
- **Tags**: sapiens, sapiens2, vision-transformer, human-centric, pretrained-backbone, feature-extraction, 4k
- **Downloads**: 10
- **Stars**: 0
- **Last updated**: 2026-10-01

Source: https://www.modelscope.cn/models/facebook/sapiens2-pretrain-1b-4k

---

# Sapiens2-1B-4K

Sapiens2 is a family of high-resolution vision transformers pretrained on **1 billion human images** — designed for human-centric tasks such as pose estimation, body-part segmentation, surface normals, and pointmaps.

This repository contains the **1B parameter pretrained backbone, trained at 4K resolution (4096 × 3072, H × W)** with a window-tokenizer front-end for tractable token counts. It produces dense per-patch features at 4K input.

- 📄 **Paper:** [arXiv:2604.21681](https://arxiv.org/pdf/2604.21681)
- 🌐 **Project Page:** [rawalkhirodkar.github.io/sapiens2](https://rawalkhirodkar.github.io/sapiens2)
- 💻 **Code:** [github.com/facebookresearch/sapiens2](https://github.com/facebookresearch/sapiens2)

## Model Details

- **Developed by:** Meta
- **Model type:** Vision Transformer
- **License:** [Sapiens2 License](https://github.com/facebookresearch/sapiens2/blob/main/LICENSE.md)
- **Task:** pretrain (4K resolution)
- **Format:** safetensors
- **File:** `sapiens2_1b_4k_pretrain.safetensors`

## Quick Start

Install the [Sapiens2 repo](https://github.com/facebookresearch/sapiens2) (`pip install -e .`).

> **Important:** This is the **4K** variant. You **must** instantiate with `use_tokenizer=True` and pass an input tensor of shape `(B, 3, 4096, 3072)` (H × W).

```python
import torch
from huggingface_hub import hf_hub_download
from safetensors.torch import load_file
from sapiens.backbones.standalone.sapiens2 import Sapiens2

# Build the model and load the 4K pretrained checkpoint
model = Sapiens2(
    arch="sapiens2_1b",
    img_size=(4096, 3072),   # H × W
    patch_size=16,
    use_tokenizer=True,      # required for the 4K variant
).eval().cuda()

ckpt_path = hf_hub_download(
    repo_id="facebook/sapiens2-pretrain-1b-4k",
    filename="sapiens2_1b_4k_pretrain.safetensors",
)
model.load_state_dict(load_file(ckpt_path))

# Forward pass on a single 4K image (RGB; ImageNet normalization recommended)
x = torch.randn(1, 3, 4096, 3072).cuda()
with torch.no_grad():
    features = model(x)[0]  # dense backbone features
```

## Model Card

| Field | Value |
|-------|-------|
| Architecture | Sapiens2 ViT (RoPE, GQA, SwiGLU, RMSNorm, QK-norm) + window tokenizer |
| Backbone parameters | 1.607 B |
| Embedding dim | 1536 |
| Layers | 40 |
| Attention heads | 24 |
| Pretraining resolution | **4096 × 3072 (H × W)** |
| Patch size | 16 |
| Window size | 4 |
| Pretraining data | 1B human images |

### Sapiens2 Family

| Model | Params | FLOPs | Embed dim | Layers | Heads |
|-------|--------|-------|-----------|--------|-------|
| [Sapiens2-0.1B](https://huggingface.co/facebook/sapiens2-pretrain-0.1b) | 0.114 B | 0.342 T | 768 | 12 | 12 |
| [Sapiens2-0.4B](https://huggingface.co/facebook/sapiens2-pretrain-0.4b) | 0.398 B | 1.260 T | 1024 | 24 | 16 |
| [Sapiens2-0.8B](https://huggingface.co/facebook/sapiens2-pretrain-0.8b) | 0.818 B | 2.592 T | 1280 | 32 | 16 |
| [Sapiens2-1B](https://huggingface.co/facebook/sapiens2-pretrain-1b) | 1.462 B | 4.715 T | 1536 | 40 | 24 |
| **Sapiens2-1B-4K** *(this)* | 1.607 B | — | 1536 | 40 | 24 |
| [Sapiens2-5B](https://huggingface.co/facebook/sapiens2-pretrain-5b) | 5.071 B | 15.722 T | 2432 | 56 | 32 |

See the [Sapiens2 Collection](https://huggingface.co/collections/facebook/sapiens2) for all variants and downstream task checkpoints (pose, segmentation, normals, pointmaps).

## Intended Use

- Feature extraction for human-centric downstream tasks at 4K resolution
- Initialization for fine-tuning high-resolution task heads (pose, segmentation, normals, pointmap)
- Research on human-centric vision at high resolution

## License

Released under the [Sapiens2 License](https://github.com/facebookresearch/sapiens2/blob/main/LICENSE.md).

## Citation

```bibtex
@article{khirodkarsapiens2,
  title={Sapiens2},
  author={Khirodkar, Rawal and Wen, He and Martinez, Julieta and Dong, Yuan and Su, Zhaoen and Saito, Shunsuke},
  journal={arXiv preprint arXiv:2604.21681},
  year={2026}
}
```