Open Arsenal

A programmable privacy layer for Earth data.

Atlas Privacy Layer is an open-source geospatial privacy and rendering framework designed to give developers fine-grained control over how Earth data is displayed, processed, and shared. Instead of relying on static imagery as provided by external mapping services, Atlas introduces a programmable intermediary layer that intercepts map tiles and applies privacy rules in real time. This allows sensitive locations to be protected through configurable transformations while still maintaining usable geographic context.

At its core, Atlas operates as a tile-level privacy proxy built for modern GIS workflows. It integrates with standard geospatial systems such as PostGIS, GDAL pipelines, and XYZ tile services, enabling seamless adoption within existing mapping infrastructures. Users can define geofenced privacy zones using polygons, buffers, or imported geographic datasets, which are then used to determine how and where transformations are applied across incoming satellite or map imagery.

The framework supports multiple privacy transformation modes, including masking, pixelation, tile scrambling, and edge degradation. For more advanced use cases, Atlas can replace real-world structures with procedurally generated or AI-inpainted synthetic terrain, effectively abstracting sensitive features while preserving the overall geographic layout. This allows maps to remain visually coherent while removing identifiable detail from protected regions.

Atlas also includes an AI Object Detection Layer that enhances its privacy capabilities by identifying features such as buildings, vehicles, infrastructure, and land-use patterns before rendering decisions are made. This enables context-aware transformations, where different object types can be handled using different privacy strategies instead of applying a uniform filter across all imagery.

To ensure extensibility, Atlas provides a plugin system that allows developers to introduce custom masking algorithms, synthetic terrain generators, AI models, and external data sources without modifying the core engine. This modular design makes it suitable for research, experimentation, and production deployments in privacy-focused geospatial applications.

Overall, Atlas Privacy Layer is designed as a programmable infrastructure component for Earth data privacy—bridging GIS systems, AI vision models, and rendering pipelines into a single, self-hosted framework that prioritizes user-controlled visibility of spatial information.