GeoMultiTaskNet: remote sensing unsupervised domain adaptation using geographical coordinates

TL;DR

This paper introduces GeoMultiTaskNet, a lightweight unsupervised domain adaptation model for remote sensing semantic segmentation that leverages geographical coordinates and class frequency sampling to improve accuracy and efficiency.

Contribution

The paper presents the first use of geographical metadata for UDA in remote sensing, combining a novel GeoMultiTask module with a dynamic class sampling strategy.

Findings

Achieves 47.22% mIoU on FLAIR dataset

Reduces model parameters to 33 million

Sets new state-of-the-art performance in RS UDA

Abstract

Land cover maps are a pivotal element in a wide range of Earth Observation (EO) applications. However, annotating large datasets to develop supervised systems for remote sensing (RS) semantic segmentation is costly and time-consuming. Unsupervised Domain Adaption (UDA) could tackle these issues by adapting a model trained on a source domain, where labels are available, to a target domain, without annotations. UDA, while gaining importance in computer vision, is still under-investigated in RS. Thus, we propose a new lightweight model, GeoMultiTaskNet, based on two contributions: a GeoMultiTask module (GeoMT), which utilizes geographical coordinates to align the source and target domains, and a Dynamic Class Sampling (DCS) strategy, to adapt the semantic segmentation loss to the frequency of classes. This approach is the first to use geographical metadata for UDA in semantic segmentation. It reaches state-of-the-art performances (47,22% mIoU), reducing at the same time the number of parameters (33M), on a subset of the FLAIR dataset, a recently proposed dataset properly shaped for RS UDA, used for the first time ever for research scopes here.