Cross Modal Distillation for Flood Extent Mapping

TL;DR

This paper introduces a cross-modal distillation approach that leverages paired multi-spectral and SAR imagery to improve flood extent mapping, reducing labeling needs and outperforming baseline models on a benchmark dataset.

Contribution

It proposes a novel cross modal distillation framework transferring supervision from multi-spectral to SAR imagery for flood detection.

Findings

Outperforms baseline by 6.53% IoU on Sen1Floods11 dataset.

Utilizes unlabelled paired data to reduce labeling requirements.

Demonstrates effectiveness of cross modal knowledge transfer in flood mapping.

Abstract

The increasing intensity and frequency of floods is one of the many consequences of our changing climate. In this work, we explore ML techniques that improve the flood detection module of an operational early flood warning system. Our method exploits an unlabelled dataset of paired multi-spectral and Synthetic Aperture Radar (SAR) imagery to reduce the labeling requirements of a purely supervised learning method. Prior works have used unlabelled data by creating weak labels out of them. However, from our experiments we noticed that such a model still ends up learning the label mistakes in those weak labels. Motivated by knowledge distillation and semi supervised learning, we explore the use of a teacher to train a student with the help of a small hand labelled dataset and a large unlabelled dataset. Unlike the conventional self distillation setup, we propose a cross modal distillation framework that transfers supervision from a teacher trained on richer modality (multi-spectral images) to a student model trained on SAR imagery. The trained models are then tested on the Sen1Floods11 dataset. Our model outperforms the Sen1Floods11 baseline model trained on the weak labeled SAR imagery by an absolute margin of 6.53% Intersection-over-Union (IoU) on the test split.