Scene-to-Patch Earth Observation: Multiple Instance Learning for Land Cover Classification
This work addresses the bottleneck of dataset creation for land cover classification, making it more accessible for climate change mitigation efforts in technology, government, and academia.
The paper tackles the problem of land cover classification by proposing Scene-to-Patch models using Multiple Instance Learning, which requires only scene-level labels instead of fully-annotated datasets, and it outperforms non-MIL baselines on the DeepGlobe-LCC dataset.
Land cover classification (LCC), and monitoring how land use changes over time, is an important process in climate change mitigation and adaptation. Existing approaches that use machine learning with Earth observation data for LCC rely on fully-annotated and segmented datasets. Creating these datasets requires a large amount of effort, and a lack of suitable datasets has become an obstacle in scaling the use of LCC. In this study, we propose Scene-to-Patch models: an alternative LCC approach utilising Multiple Instance Learning (MIL) that requires only high-level scene labels. This enables much faster development of new datasets whilst still providing segmentation through patch-level predictions, ultimately increasing the accessibility of using LCC for different scenarios. On the DeepGlobe-LCC dataset, our approach outperforms non-MIL baselines on both scene- and patch-level prediction. This work provides the foundation for expanding the use of LCC in climate change mitigation methods for technology, government, and academia.