Official Pytorch code of "A Deep Multi-task Convolutional Neural Network for Remote Sensing Image Super-resolution and Colorization", Accepted by IEEE Transactions on Geoscience and Remote Sensing (TGRS) 2022 [PDF]
Remote sensing data have become increasingly vital in target detection, disaster monitoring, and military surveillance. Abundant pan-sharpening and super-resolution (SR) methods based on deep learning have been proposed and have achieved remarkable performance. However, pan-sharpening requires paired panchromatic (PAN) and multispectral (MS) images, and SR cannot increase the spectral resolution of PAN. Thus, we introduce a computational imaging-based method to recover or produce the incomplete data of single PAN or MS. This work also explores the integration of multiple tasks by a single neural network. We start with SR and colorization, study the feasibility of simultaneously finishing SR colorization, and use a model trained in SR colorization to finish pan-sharpening without MS. A generic neural network, RSI-Net, is designed for remote sensing image SR, colorization, simultaneous SR colorization, and pan-sharpening. To verify its performance, RSI-Net is compared with state-of-the-art SR and colorization methods. Experiments show that RSI-Net can be competitive in visual effects and evaluation indexes, and it performs well at simultaneous SR colorization, and RSI-Net finishes pan-sharpening only need to input PAN. Our experiments confirm the effect of integrating multiple tasks.
- Python 3.8
- torch == 1.10.0+cu113
- torchvision == 0.11.0+cu113
- numpy
- skimage
- cv2
- tqdm
- pytorch_msssim
​ 1.1 Download the NWPU-45 dataset for training, AID dataset, and RSSCN7 dataset for testing.
​ 1.2 Run prepare.py to prepare the train and test dataset(NWPU-45). Please Update root_path and save_path .
python prepare.py --root_path XXX --save_path XXX​ Examples:
python prepare.py --root_path /home/eric/PycharmProjects/NWPU-RESISC45/ --save_path ./data/​ 2.1 Run train.py to train the model. Please update the task , img_train_path, img_test_path.
python train.py --task XXX --img_train_path XXX --img_test_path XXX​ Examples:
python train.py --task srx2 --img_train_path ./data/train/label/ --img_test_path ./data/test/label/Pre-trained Models:
| URL | Extract Code |
|---|---|
| https://pan.baidu.com/s/1Cpaqb2wmIIAfDGu-XxI5NA | tgrs |
Run test.py to test RSINet on Colorzation, SRX2, SRX4, SRX8, and Colorization SRX2 tasks.:
​ For colorization task:
python test.py --task color --best_pkl_path ./save_model/color.pkl​ For SR X2 task:
python test.py --task srx2 --best_pkl_path ./save_model/srx2.pkl​ For SR X4 task:
python test.py --task srx4 --best_pkl_path ./save_model/srx4.pkl​ For SR X8 task:
python test.py --task srx8 --best_pkl_path ./save_model/srx8.pkl​ For Colorzation and SR X2 task:
python test.py --task colorx2 --best_pkl_path ./save_model/colorx2.pkl@ARTICLE{9721252,
author={Feng, Jianan and Jiang, Qian and Tseng, Ching-Hsun and Jin, Xin and Liu, Ling and Zhou, Wei and Yao, Shaowen},
journal={IEEE Transactions on Geoscience and Remote Sensing},
title={A Deep Multitask Convolutional Neural Network for Remote Sensing Image Super-resolution and Colorization},
year={2022},
volume={},
number={},
pages={1-1},
doi={10.1109/TGRS.2022.3154435}}
