DSFN

KO | EN

gitlite — search

#llm #go #machine-learning #rust #linux #docker #macos #java #windows #cli #kubernetes #pytorch #security #open-source

(★ 17)

NeurIPS 2025 -- Depth-Supervised Fusion Network for Seamless-Free Image Stitching

DSFN 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Depth-Supervised Fusion Network for Seamless-Free Image Stitching [NeurIPS 2025] [[*Arxiv*]](https://arxiv.org/abs/2510.21396)

Zhiying Jiang1 &ensp; Ruhao Yan2 &ensp; Zengxi Zhang2 &ensp; Bowei Zhang2 &ensp; Jinyuan Liu2

1College of Information Science and Technology, Dalian Maritime University

2School of Software Technology, Dalian University of Technology

## Updates
[2025-09-18] Our paper has been accepted by NeurIPS 2025!

[2025-10-09] The code and the pre-trained model are available.

[2025-11-09] The Chinese translation of our paper is available. [[Chinese translation version]](./Paper/NeurIPS2025_CN.pdf)

## Code
**Requirements**

- Linux
- Python 3
- numpy >= 1.19.5
- pytorch >= 1.7.1
- CUDA >= 11.0

We use an NVIDIA RTX 3090 GPU to achieve this task. If you are using hardware of other models, please adapt the versions of tools such as CUDA and pytorch by yourself.

## Data preparation

Please download UDIS-D dataset([Google Drive](https://drive.google.com/drive/folders/1kC7KAULd5mZsqaWnY3-rSbQLaZ7LujTY) or [Baidu Yun](https://pan.baidu.com/share/init?surl=3KZ29e487datgtMgmb9laQ?pwd=1234)) and IVSD dataset([Google Drive](https://drive.google.com/file/d/1EFS0O-3KujvRJvcRx_Me5W2fdn9jRKGc) or [Baidu Yun](https://pan.baidu.com/share/init?surl=ZP4hgBovXnsLHcOReCGnrg&pwd=ssfv)) for training and testing.

Please first use [DepthAnything](https://github.com/DepthAnything/Depth-Anything-V2/tree/main) to generate the depth map of the dataset.

**pre-trained models**

Please download the pre-trained model([Google Drive](https://drive.google.com/drive/folders/1qsvIj7iN62gdHJLytTAzfvrOZaGaqXzR) or [Baidu Yun](https://pan.baidu.com/s/1VpBGxePAbQoL7IbNspQMuA?pwd=2025)) and place it in the corresponding path.

## How to Run

**Installation**

You can create a Conda environment through the following command:
```
conda env create -f environment.yml
```

**Training**

Please set the training dataset path in Warp/newCodes/train.py and Fusion/newCodes/train.py, and run the following command under the corresponding path to start the training.

```
python train.py
```
***Please note that the training in the fusion phase depends on the result of the warp phase. Pay attention to the training sequence. After completing the training in the warp phase, warp the training set to generate the fusion training set.***

**Testing**

Please set the training dataset path in Warp/newCodes/test_output.py and Fusion/newCodes/test.py, and run the following command under the corresponding path to start the training.

>For Warp:
>```
>python test_output.py
>```
>For Fusion:
>```
>python test.py
>```

***Please note that, similar to the training phase, the testing phase still has sequential nature.***

## Experimental Results

**Visual comparison of stitched images from UDIS-D dataset and IVSD dataset.**

**Quantitative comparison on UDIS-D and IVSD datasets.**

## Any Question
If you have any other questions about the code, please email: yanruhao1997@hotmail.com

## Citation
If you think this work is helpful to you, you can cite this paper!
```
@misc{jiang2025depthsupervisedfusionnetworkseamlessfree,
      title={Depth-Supervised Fusion Network for Seamless-Free Image Stitching}, 
      author={Zhiying Jiang and Ruhao Yan and Zengxi Zhang and Bowei Zhang and Jinyuan Liu},
      year={2025},
      eprint={2510.21396},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2510.21396}, 
}

@inproceedings{jiangdepth,
  title={Depth-Supervised Fusion Network for Seamless-Free Image Stitching},
  author={Jiang, Zhiying and Yan, Ruhao and Zhang, Zengxi and Zhang, Bowei and Liu, Jinyuan},
  booktitle={The Thirty-ninth Annual Conference on Neural Information Processing Systems}
}
```

DSFN

DSFN 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

DSFN

DSFN 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README