Meta-Learning for Few-Shot Land Cover Classification

TL;DR

This paper explores using meta-learning, specifically MAML, to improve land cover classification across diverse geographic regions with limited data, outperforming traditional methods in cross-region adaptation.

Contribution

It demonstrates the effectiveness of meta-learning for few-shot adaptation in Earth surface classification, addressing regional diversity challenges.

Findings

Meta-learning outperforms pre-training in cross-region tasks

Few-shot adaptation achieves better results with limited data

Traditional supervised learning remains effective without domain shift

Abstract

The representations of the Earth's surface vary from one geographic region to another. For instance, the appearance of urban areas differs between continents, and seasonality influences the appearance of vegetation. To capture the diversity within a single category, like as urban or vegetation, requires a large model capacity and, consequently, large datasets. In this work, we propose a different perspective and view this diversity as an inductive transfer learning problem where few data samples from one region allow a model to adapt to an unseen region. We evaluate the model-agnostic meta-learning (MAML) algorithm on classification and segmentation tasks using globally and regionally distributed datasets. We find that few-shot model adaptation outperforms pre-training with regular gradient descent and fine-tuning on (1) the Sen12MS dataset and (2) DeepGlobe data when the source domain and target domain differ. This indicates that model optimization with meta-learning may benefit tasks in the Earth sciences whose data show a high degree of diversity from region to region, while traditional gradient-based supervised learning remains suitable in the absence of a feature or label shift.