Satim offers OREC, an AI‑driven Automatic Target Recognition SaaS that ingests raw Synthetic Aperture Radar data from any sensor and delivers near‑real‑time detection and classification of ships, aircraft, and ground vehicles. The platform uses three specialized convolutional neural networks and a proprietary SAR data simulator to expand object classes without retraining, and provides results through a secure cloud dashboard and RESTful APIs for integration with command‑and‑control systems.
Funding
Funding not disclosed
Founders
Product
Problem
Analyzing Synthetic Aperture Radar (SAR) imagery for object detection and classification is traditionally labor‑intensive and relies on bespoke processing pipelines, which limits timely situational awareness for defense, government, and commercial users.
Solution
Satim delivers OREC, an AI‑driven Automatic Target Recognition (ATR) platform that processes SAR data in near‑real‑time to identify and classify ships, aircraft, and ground vehicles. The system employs three dedicated deep‑learning models that are data‑agnostic, enabling consistent performance across different SAR sensors and acquisition modes. Results are streamed to a secure cloud service where they are indexed, visualized, and made available via API for downstream analytics. A proprietary SAR data simulator accelerates the creation of new object classes, allowing customers to extend detection capabilities as mission requirements evolve. The solution is packaged as a SaaS offering with optional on‑premise deployment for highly classified environments.
Target Audience
Primary customers are defense and intelligence agencies, maritime surveillance operators, and commercial remote‑sensing firms that require automated, high‑precision SAR object detection for operational planning and threat monitoring.
Features
- Three high‑accuracy convolutional neural network models specialized for maritime, airborne, and vehicular targets in SAR imagery
- Near‑real‑time inference pipeline (≤ 2 seconds per scene) with GPU‑accelerated processing
- Data‑agnostic architecture that ingests raw SAR products from any vendor without re‑training
- Proprietary SAR data simulator that generates labeled synthetic scenes for rapid model expansion to new object categories
- RESTful API and SDKs (Python, C++) for seamless integration with existing command‑and‑control or analytics platforms
- Cloud‑native analytics dashboard with heat‑maps, confidence scores, and export to standard geospatial formats (GeoJSON, KML)
- End‑to‑end encryption and role‑based access control to meet classified and GDPR compliance requirements