README.md

---
language:
- en
license: bsd-3-clause
tags:
- arxiv:2506.23151
- optical-flow-estimation
pipeline_tag: image-to-image
library_name: pytorch
base_model:
- egorchistov/MEMFOF-Tartan-T
---

# MEMFOF-Tartan-T-TSKH

<a href="https://arxiv.org/abs/2506.23151" style="text-decoration: none;">📄 Paper</a> | <a href="https://msu-video-group.github.io/memfof" style="text-decoration: none;">🌐 Project Page</a> | <a href="https://github.com/msu-video-group/memfof" style="text-decoration: none;">💻 Code</a> | <a href="https://colab.research.google.com/github/msu-video-group/memfof/blob/dev/demo.ipynb" style="text-decoration: none;">🚀 Colab</a> | <a href="https://huggingface.co/spaces/egorchistov/MEMFOF" style="text-decoration: none;">🤗 Demo</a>

🔍 This is a MEMFOF checkpoint trained from **MEMFOF-Tartan-T** on the combination of **FlyingThings3D, Sintel, KITTI, and HD1K** datasets.

✅ **Note:** This model is intended **for real-world videos** — it is trained with **higher diversity and robustness** in mind.

## 🛠️ Usage

```shell
git clone https://github.com/msu-video-group/memfof.git
cd memfof
pip3 install -r requirements.txt
```

```python
import torch
from core.memfof import MEMFOF

device = "cuda" if torch.cuda.is_available() else "cpu"
model = MEMFOF.from_pretrained("egorchistov/MEMFOF-Tartan-T-TSKH").eval().to(device)

with torch.inference_mode():
    example_input = torch.randint(0, 256, [1, 3, 3, 1080, 1920], device=device)  # [B=1, T=3, C=3, H=1080, W=1920]
    backward_flow, forward_flow = model(example_input)["flow"][-1].unbind(dim=1)  # [B=1, C=2, H=1080, W=1920]
```

## 📚 Citation

```
@article{bargatin2025memfof,
  title={MEMFOF: High-Resolution Training for Memory-Efficient Multi-Frame Optical Flow Estimation},
  author={Bargatin, Vladislav and Chistov, Egor and Yakovenko, Alexander and Vatolin, Dmitriy},
  journal={arXiv preprint arXiv:2506.23151},
  year={2025}
}
```