Feature guided masked Autoencoder for self-supervised learning in remote sensing
PyTorch implementation of FGMAE.
| modality | ViT-Small | ViT-Base | ViT-Large | ViT-Huge |
|---|---|---|---|---|
| 13 band MS | ViT-S/16 | ViT-B/16 | ViT-L/16 | ViT-H/14 |
| 2 band SAR | ViT-S/16 | ViT-B/16 | ViT-L/16 | ViT-H/14 |
For FGMAE-MS, go to src/pretrain_ssl, and run
python pretrain_fgmae.py \
--root_dir /path/to/data \
--output_dir /path/to/checkpoints \
--log_dir /path/to/logs \
--model mae_vit_small_patch16 \
--mask_ratio 0.7 \
--num_workers 10 \
--batch_size 64 \
--epochs 100 \
--warmup_epochs 10 \
--input_size 224 \
--feature hog+ndi \
--blr 1.5e-4 \
--in_channels 13 \
--norm_pix_loss \
--hog_norm \
For FGMAE-SAR, modify --in_channels 2 and --feature hog.
On one node in a slurm system with 4 GPUs, we provide some example job submission scripts in src/scripts.
See src/pretrain_ssl/transfer_classification for linear probing and fine tuning on BigEarthNet and EuroSAT.
We collect a Sentinel-1 GSD SAR version of EuroSAT by matching the coordinates. Please see EuroSAT-SAR for descriptions and downloads.
This repository is heavily built on MAE, which is under the Apache 2.0 License - see the LICENSE file for details.
@article{wang2023feature,
title={Feature Guided Masked Autoencoder for Self-supervised Learning in Remote Sensing},
author={Wang, Yi and Hern{\'a}ndez, Hugo Hern{\'a}ndez and Albrecht, Conrad M and Zhu, Xiao Xiang},
journal={arXiv preprint arXiv:2310.18653},
year={2023}
}