land-cover

Code for "Lucas Hu, Caleb Robinson, and Bistra Dilkina. 2021. Model Generalization in Deep Learning Applications for Land Cover Mapping. In Fragile Earth ’21: KDD 2021 Workshop. ACM, New York, NY, USA, 9 pages." (https://arxiv.org/abs/2008.10351)

Setup Instructions

1. git clone this repository
2. cd into root of this repository
3. pip install -r requirements.txt
4. git clone https://github.com/kobiso/CBAM-keras
5. git clone https://github.com/titu1994/DenseNet
6. Download and extract SEN12MS dataset: https://arxiv.org/abs/1906.07789
7. Edit dataset paths in config.json
8. Run one_time_scripts/save_segmentation_patches_to_npy.py to preprocess dataset

Usage Instructions

• Train models: python classify.py -c config.json --train
• Generate predictions: python classify.py -c config.json --predict

Sentinel-2 Band Distributions (By Continent)

Sentinel-2 Band Distributions (By Season)