Foundation Models

Let's explore the incredible power of self supervised methodology applied to earth observation

Pleaides Neo Foundation Model

At Disaitek, we leverage the state-of-the-art Foundation Model, trained with 6 bands and 30cm imagery from the Pleiades Neo satellites using advanced self-supervised learning methods. This innovative approach drastically reduces the dependence on extensive labeled datasets, enabling faster and more cost-effective application deployment

How can we meet the challenge of building applications that use the features of the Pleaides Neo satellites without having access to large datasets to train classification models?

Key Benefits

High Spatial Resolution (30cm): Enables precise identification of land use land cover and subtle environmental changes.
Rapid Deployment & Adaptation: Quickly applicable to diverse environmental or anthrpogenic scenarios with minimal effort.
Significant Reduction in Labeling Costs: Fine tuning the Foundation Model minimizes the need for extensive manual annotations.

Applications

Urban & Infrastructure: Building footprints, structures under construction, road networks, mobility patterns, density of vehicles (personal, commercial, trucks, buses) per grid cell
Commercial & Real Estate: Retail footprints, competitor locations, parking availability, accessibility to transit, property condition indicators
Ecological & Green: Forested areas, shrubs, hedgerows, isolated trees, ecological corridors, vegetation health indices (e.g., NIR-based), green coverage vs. impervious surfaces, detection of illegal logging and habitat disturbances
Land Parcel Integrity: Detect unauthorized structures, informal settlements, and boundary encroachments within legally recognized property parcels.
Encroachment Indicators: Monitor clearing of vegetation, new pathways, or concrete wall signaling illegal land subdivision.
Illegal End-of-Use Vehicles (ELV) Activity: Identify clusters of abandoned or end-of-use vehicles, unlicensed scrapyards, and large-scale vehicle dumping sites.
Environmental Damage Markers: Recognize patterns of soil degradation or material stockpiles indicative of illegal vehicle dismantling or hazardous waste disposal.
Temporal Changes: Compare past and current imagery to detect incremental encroachment, evolving ELV hotspots, and seasonal patterns in illicit activities.

Anchor 1

Discover How Our Foundation Model Can Transform Your Applications