This is the official PyTorch implementation of the MAGE paper (TGRS).

All criticisms, suggestions, questions are welcome.

• Refactoring
• A Docker containing all environments

• Dependency
  • See requirements.txt
• Dataset
  • baiduwangpan, extraction code: d7x3
  • Google Drive
• Checkpoints
  • baiduwangpan, extraction code: safz
  • Google Drive

After configuring the environment and replacing the dataset and checkpoint folders with contents from the above links, one can use the following commands to use checkpoints for evaluation or train from scratch.

• Evaluation with checkpoints

python test.py

• Train from scratch

python main.py

Note: Relevant work should be cited when using the dataset to avoid copyright disputes.

• MUUFL
• Trento
• Houston

• FusAtNet: Dual Attention Based SpectroSpatial Multimodal Fusion Network for Hyperspectral and LiDAR classification
• S²ENet: Spatial–Spectral Cross-Modal Enhancement Network for Classification of Hyperspectral and LiDAR Data
• Deep Encoder–Decoder Networks for Classification of Hyperspectral and LiDAR Data
• More diverse means better: Multimodal deep learning meets remote-sensing imagery classification
• Learning from labeled and unlabeled data with label propagation

If you find this code to be useful for your research, please consider citing.

@ARTICLE{9904571,
  author={Xiu, Di and Pan, Zongxu and Wu, Yirong and Hu, Yuxin},
  journal={IEEE Transactions on Geoscience and Remote Sensing}, 
  title={MAGE: Multisource Attention Network with Discriminative Graph and Informative Entities for Classification of Hyperspectral and LiDAR Data}, 
  year={2022},
  volume={},
  number={},
  pages={1-1},
  doi={10.1109/TGRS.2022.3210398}}

• NVIDIA/DALI - A GPU-accelerated library containing highly optimized building blocks and an execution engine for data processing to accelerate deep learning training and inference applications.
• pyg-team/pytorch_geometric - Graph Neural Network Library for PyTorch
• wandb - A tool for visualizing and tracking your machine learning experiments.