Semantically-consistent Landsat 8 image to Sentinel-2 image translation for alpine areas

TL;DR

This paper explores domain adaptation via style transfer to improve segmentation accuracy between Landsat 8 and Sentinel-2 satellite images in alpine regions, demonstrating significant performance gains.

Contribution

It introduces the use of the HRSemI2I style transfer model for sensor discrepancy reduction and compares different generalization schemes for land cover classification.

Findings

Significant improvement in segmentation performance with domain-adapted images.

Effective application of style transfer to 6-band satellite imagery.

Validation of generalization schemes across different land cover label sets.

Abstract

The availability of frequent and cost-free satellite images is in growing demand in the research world. Such satellite constellations as Landsat 8 and Sentinel-2 provide a massive amount of valuable data daily. However, the discrepancy in the sensors' characteristics of these satellites makes it senseless to use a segmentation model trained on either dataset and applied to another, which is why domain adaptation techniques have recently become an active research area in remote sensing. In this paper, an experiment of domain adaptation through style-transferring is conducted using the HRSemI2I model to narrow the sensor discrepancy between Landsat 8 and Sentinel-2. This paper's main contribution is analyzing the expediency of that approach by comparing the results of segmentation using domain-adapted images with those without adaptation. The HRSemI2I model, adjusted to work with 6-band imagery, shows significant intersection-over-union performance improvement for both mean and per class metrics. A second contribution is providing different schemes of generalization between two label schemes - NALCMS 2015 and CORINE. The first scheme is standardization through higher-level land cover classes, and the second is through harmonization validation in the field.