Activation Regression for Continuous Domain Generalization with Applications to Crop Classification

TL;DR

This paper addresses geographic variance in satellite imagery by modeling it as a continuous domain adaptation problem, improving crop classification across the US using domain knowledge and feature regression.

Contribution

It introduces a novel approach combining domain knowledge with feature regression in satellite imagery to enhance geographic generalisation for crop classification.

Findings

Improved generalisation by passing climate variables to a Transformer model.

Enhanced model performance through feature regression of domain variables.

Provided a new dataset covering the entire US for crop classification research.

Abstract

Geographic variance in satellite imagery impacts the ability of machine learning models to generalise to new regions. In this paper, we model geographic generalisation in medium resolution Landsat-8 satellite imagery as a continuous domain adaptation problem, demonstrating how models generalise better with appropriate domain knowledge. We develop a dataset spatially distributed across the entire continental United States, providing macroscopic insight into the effects of geography on crop classification in multi-spectral and temporally distributed satellite imagery. Our method demonstrates improved generalisability from 1) passing geographically correlated climate variables along with the satellite data to a Transformer model and 2) regressing on the model features to reconstruct these domain variables. Combined, we provide a novel perspective on geographic generalisation in satellite imagery and a simple-yet-effective approach to leverage domain knowledge. Code is available at: \url{https://github.com/samar-khanna/cropmap}