UDL

UDL is a practicable framework used in Deep Learning (computer vision).

Benchmark

codes, results and models are available in UDL, please contact @Liang-Jian Deng (corresponding author)

Results of DCFNet

Quantitative results

Visual results

please see the paper and the sub-directory: ./UDL/results/DCFNet

Install [Option]

please run python setup.py develop

Usage

open UDL/panshaprening/tests, run the following code:

python run_DCFNet.py

Note that default configures don't fit other environments, you can modify configures in pansharpening/models/DCFNet/option_DCFNet.py.

Benefit from mmcv/config.py, the project has the global configures in Basis/option.py, option_DCFNet inherits directly from Basis/option.py.

1. Data preparation

You need to download WorldView-3 datasets.

The directory tree should be look like this:

|-$ROOT/datasets
├── pansharpening
│   ├── training_data
│   │   ├── train_wv3_10000.h5
│   │   ├── valid_wv3_10000.h5
│   ├── test_data
│   │   ├── WV3_Simu
│   │   │   ├── new_data6.mat
│   │   │   ├── new_data7.mat
│   │   │   ├── ...
│   │   ├── WV3_Simu_mulExm
│   │   │   ├── test1_mulExm1258.mat

2. Training

args.eval = False, args.dataset='wv3'

3. Inference

args.eval = True, args.dataset='wv3_singleMat'

Plannings

Please expect more tasks and models

• pansharpening

  • models

• derain

  • models

• HISR

  • models

Contribution

We appreciate all contributions to improve UDL. Looking forward to your contribution to UDL.

Citation

If you use this toolbox or benchmark in your research, please cite this project.

@InProceedings{Wu_2021_ICCV,
    author    = {Wu, Xiao and Huang, Ting-Zhu and Deng, Liang-Jian and Zhang, Tian-Jing},
    title     = {Dynamic Cross Feature Fusion for Remote Sensing Pansharpening},
    booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
    month     = {October},
    year      = {2021},
    pages     = {14687-14696}
}

Acknowledgement

• MMCV: OpenMMLab foundational library for computer vision.
• HRNet : High-resolution networks and Segmentation Transformer for Semantic Segmentation

License & Copyright

This project is open sourced under GNU General Public License v3.0