Adaptive-attention Completing Network (AACNet) for Remote Sensing Image

The code will be available.

This is the testing code for our AACNet to reconstruct remote sensing and natural image. Given one image and mask, the proposed **AACNet** model is able to reconstruct masked regions. This code is adapted from an initial fork of [PIC](https://github.com/lyndonzheng/Pluralistic-Inpainting) implementation.

Illustration of Ada-attention

Illustration of self-attention and Ada-attention, including high attention score points, key/value positions, and inpainting results. The gray covering regions represent the corrupted regions. In the high attention score points images, the red stars show the specific query, and the purple circles show the high attention score points. These attention score images show that Ada-attention focuses more on relevant keys related to the query, e.g., the roof. In the key/value positions images, cyan dots denote the original uniform coordinates used in self-attention, and red dots of Ada-attention denote the sampled coordinates adjusted by the offset subnet, which are more inclined to the edge and texture with rich features. In the inpainting results images, local details are displayed in the red box, demonstrating that our Ada-attention generates superior results.

Getting started

Installation

This code was tested with Pytoch 1.8.1 CUDA 11.1, Python 3.6 and Ubuntu 18.04

• Create conda environment:

conda create -n inpainting-py36 python=3.6
conda deactivate
conda activate inpainting-py36
pip install visdom dominate

• Clone this repo:

git clone https://github.com/huangwenwenlili/AACNet
cd AACNet

• Pip install libs:

pip install -r requirements.txt

Datasets

• AID: It contains 30 scene categories of RS aerial RGB images. 10000 images. AID can download on Onedrive or BaiduPan.
• PatternNet: It contains 45 scene categories of RS digital images. 30400 images. PatternNet
• NWPU-RESISC45: It contains 38 scene categories of RS digital images. 31500 images. NWPU-RESISC45
• Paris StreetView: It contains buildings of Paris of natural digital images. 14900 training images and 100 testing images. Paris
• CelebA-HQ: It contains celebrity face images. 30000 images. CelebA-HQ

Testing

• Test the model. Input images and masks resolution are 256*256. In the testing, we use irregular mask dataset to evaluate different ratios of corrupted region images.

python test.py  --name aid --img_file your_image_path

• Set --mask_type in options/base_options.py to test various masks. --mask_file path is needed for 3. external irregular mask,
• The default results will be saved under the results folder. Set --results_dir for a new path to save the result.

Pretrained Models

Download the pre-trained models using the following links and put them undercheckpoints/ directory. BaiduPan link: https://pan.baidu.com/s/1OND83WeaViSUOGu7v_DJXg?pwd=skix extract code:skix

• RS model: AID | PatternNet | NWPU-RESISC45
• Natural model: Paris StreetView | CelebA-HQ

Our models are trained with images of resolution 256*256 with random regular and irregular holes.

Example Results

• Completion Results for RS Datasets

• Completion Results for Natural Inpainting Datasets

License

The codes and the pre-trained models in this repository are under the MIT license as specificed by the LICENSE file. This code is for educational and academic research purpose only.

Reference Codes

• https://github.com/lyndonzheng/Pluralistic-Inpainting
• https://github.com/NVIDIA/partialconv

Citation

If you use this code for your research, please cite our paper.

@article{huang2023adaptive,
  title={Adaptive-attention Completing Network (AACNet) for Remote Sensing Image},
  author={Huang, Wenli and Deng, Ye and Hui, Siqi and Wang, Jinjun},
  journal={},
  volume={},
  pages={},
  year={}
}