https://github.com/Ehzoahis/DEReD

The Offical Codebase for Fully Self-Supervised Depth Estimation from Defocus Clue
https://github.com/Ehzoahis/DEReD

Last synced: 3 months ago
JSON representation

The Offical Codebase for Fully Self-Supervised Depth Estimation from Defocus Clue

• Host: GitHub
• URL: https://github.com/Ehzoahis/DEReD
• Owner: Ehzoahis
• Created: 2023-02-15T07:09:53.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2023-06-03T04:48:43.000Z (over 1 year ago)
• Last Synced: 2024-05-28T13:19:35.993Z (5 months ago)
• Language: Python
• Homepage: https://ehzoahis.github.io/DEReD/
• Size: 49.9 MB
• Stars: 37
• Watchers: 2
• Forks: 7
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• Awesome-Monocular-Depth - Fully Self-Supervised Depth Estimation from Defocus Clue

README

        
# DEReD (Depth Estimation via Reconstucting Defocus Image)

[![Project Website](https://img.shields.io/badge/Project-Website-orange)](https://ehzoahis.github.io/DEReD)

[![arXiv](https://img.shields.io/badge/arXiv-2303.11791-b31b1b.svg)](https://arxiv.org/pdf/2303.10752.pdf)

Official codes of CVPR 2023 [Paper](https://arxiv.org/pdf/2303.10752.pdf) | _Fully Self-Supervised Depth Estimation from Defocus Clue_

## Prepreation

### Environment

Create a new environment and install dependencies with `requirement.txt`:

```shell

conda create -n dered

conda activate dered

conda install --file requirements.txt

python gauss_psf/setup.py install

```

### Data 

The generation code for NYUv2 Focal Stack dataset is provided.

The generation code for DefocusNet can be found [here](https://github.com/dvl-tum/defocus-net).

### Weight

You can download the model weights trained on NYUv2 Focal Stack from [here](https://drive.google.com/file/d/1LQUt7Lo6KPKb0OsBETkvxeujOVdqOGjh/view?usp=share_link).

## Usage

### Train

```shell

python scripts/train.py --data_path [path/to/dataset] --dataset [Dataset] --recon_all \ 

-N [experiment_name] --use_cuda -E 1000 --BS 32 --save_checkpoint --save_best --save_last \

--sm_loss_beta 2.5 --verbose --recon_loss_lambda 1e3 --aif_blur_loss_lambda 10 \

--blur_loss_lambda 1e1 --sm_loss_lambda 1e1 --log --vis

```

### Evaluation

```shell

python scripts/train.py --data_path [path/to/dataset] --dataset [Dataset] --recon_all \

-N [experiment_name] --use_cuda --BS 32 --save_best --verbose --eval

```

## Acknowledgement

Parts of the code are developed from [DefocusNet](https://github.com/dvl-tum/defocus-net) and [UnsupervisedDepthFromFocus](https://github.com/shirgur/UnsupervisedDepthFromFocus).

## Citation

```bibtex

@article{si2023fully,

  title={Fully Self-Supervised Depth Estimation from Defocus Clue},

  author={Si, Haozhe and Zhao, Bin and Wang, Dong and Gao, Yupeng and Chen, Mulin and Wang, Zhigang and Li, Xuelong},

  journal={arXiv preprint arXiv:2303.10752},

  year={2023}

}

```

## Contact Authors

[Haozhe Si](mailto:[email protected]), [Bin Zhao](mailto:[email protected]), [Dong Wang](mailto:[email protected]), [Xuelong Li](mailto:[email protected])