Transfer Learning for Onboard Cloud Segmentation in Thermal Earth Observation: From Landsat to a CubeSat Constellation

TL;DR

This paper demonstrates how transfer learning with lightweight neural networks can enable accurate, real-time onboard cloud segmentation for thermal Earth observation CubeSats, overcoming hardware and data limitations.

Contribution

It introduces a transfer learning approach using a pretrained UNet with MobileNet encoder for thermal cloud segmentation on CubeSats, achieving high accuracy with limited mission-specific data.

Findings

Pretraining on Landsat data improves segmentation accuracy.

Model inference runs in under 5 seconds on NVIDIA Jetson Nano.

Lightweight models enable real-time onboard processing.

Abstract

Onboard cloud segmentation is a critical yet underexplored task in thermal Earth observation (EO), particularly for CubeSat missions constrained by limited hardware and spectral information. CubeSats often rely on a single thermal band and lack sufficient labeled data, making conventional cloud masking techniques infeasible. This work addresses these challenges by applying transfer learning to thermal cloud segmentation for the FOREST-2 CubeSat, using a UNet with a lightweight MobileNet encoder. We pretrain the model on the public Landsat-7 Cloud Cover Assessment Dataset and fine-tune it with a small set of mission-specific samples in a joint-training setup, improving the macro F1 from 0.850 to 0.877 over FOREST-2-only baselines. We convert the model to a TensorRT engine and demonstrate full-image inference in under 5 seconds on an NVIDIA Jetson Nano. These results show that leveraging public datasets and lightweight architectures can enable accurate, efficient thermal-only cloud masking on-orbit, supporting real-time decision-making in data-limited EO missions.