Our satellite detection engine scans the world's oceans, rivers, and coastlines every 48 hours — identifying plastic waste accumulation zones with 94% confidence using multispectral imaging and machine learning.
Our pipeline combines multispectral satellite data with trained neural networks — verified by on-the-ground contributors to create the most accurate plastic detection dataset on Earth.
We ingest data from 6 satellite constellations including Sentinel-2 and Landsat-9. Each scan captures 13 spectral bands, giving our models far more signal than visible light alone — including near-infrared signatures unique to floating plastics.
Our convolutional neural network was trained on 2.3 million labelled satellite patches — verified by expert oceanographers. It distinguishes plastic aggregations from sea foam, algal blooms, and other natural surface features with high precision.
Every satellite detection is cross-referenced with submissions from our global network of 3,800+ contributors — citizens, NGO field workers, and research vessels — who submit photos and GPS coordinates to verify detections in real time.
New ground-truth data feeds back into our training pipeline weekly. As our contributor network grows, detection accuracy improves — creating a virtuous cycle where more participation means better data for everyone.
Every parameter of our detection pipeline is engineered for reliability — because ESG reporting and cleanup decisions depend on data you can stake your reputation on.
Our detection network covers all major ocean gyres, the 50 most polluted river systems, and 180,000km of coastline across 190 countries. Data is available via API or dashboard.
Get API access or explore our dashboard — free for NGOs and researchers.
View solutions See live campaigns