Learned Image Compression for Earth Observation: Implications for Downstream Segmentation Tasks

TL;DR

This paper evaluates learned image compression methods for Earth observation data, demonstrating their advantages over traditional codecs in large-scale optical imagery segmentation tasks, while also analyzing their limitations and the impact of joint optimization.

Contribution

It introduces a task-specific learned compression approach for EO data and compares its performance with traditional codecs across multiple segmentation tasks.

Findings

Learned compression outperforms JPEG 2000 in large-scale optical imagery segmentation.

Traditional codecs remain competitive on small, thermal infrared datasets.

Joint end-to-end optimization does not enhance segmentation performance.

Abstract

The rapid growth of data from satellite-based Earth observation (EO) systems poses significant challenges in data transmission and storage. We evaluate the potential of task-specific learned compression algorithms in this context to reduce data volumes while retaining crucial information. In detail, we compare traditional compression (JPEG 2000) versus a learned compression approach (Discretized Mixed Gaussian Likelihood) on three EO segmentation tasks: Fire, cloud, and building detection. Learned compression notably outperforms JPEG 2000 for large-scale, multi-channel optical imagery in both reconstruction quality (PSNR) and segmentation accuracy. However, traditional codecs remain competitive on smaller, single-channel thermal infrared datasets due to limited data and architectural constraints. Additionally, joint end-to-end optimization of compression and segmentation models does not improve performance over standalone optimization.