On-Board Anomaly Detection for Efficient Marine Environmental Monitoring
Self-supervised encoder plus anomaly detection pipeline compresses hyperspectral satellite imagery onboard to flag oil spills and algal blooms.
The paper proposes a marine event detection pipeline for Earth-observation satellites with multi- or hyperspectral sensors, combining a self-supervised neural encoder that compresses images into a latent space with a machine learning anomaly detector. It outperforms Isolation Forest, One-Class SVM, and Local Outlier Factor baselines while running on resource-constrained hardware from embedded CPUs to AI accelerators. The pipeline is being integrated into ESA's PhiSat-2 and the Microsoft/Thales Alenia Space IMAGIN-e missions, reducing downlink bandwidth by prioritizing transmission of critical information.
- Self-supervised encoder compresses satellite images for onboard processing
- Outperforms Isolation Forest, One-Class SVM, and Local Outlier Factor baselines
- Runs on embedded CPUs through AI hardware accelerators
- Integrated into ESA PhiSat-2 and Microsoft/Thales IMAGIN-e missions
- Tags
- anomaly-detectionself-supervised-learningearth-observationhyperspectralsatellitemarine-monitoring
From the original article
1 min read at arxiv.orgMarine ecosystems are impacted by various threats such as oil spills, algal blooms, and sediment floods, which disrupt habitats, wildlife, and human activities. Advances in satellite imagery and Artificial Intelligence (AI) have enhanced our capabilities for early detection and mitigation of such hazards.
By Thomas Goudemant · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.