CorteX-AI Brian Sample efficient deep learning & inference system

OptiX is UnitX’s software-defined optical imaging system that replaces trial-and-error, bespoke lighting with precise, software-controlled illumination. By dynamically shaping patterns and control lights from angles, OptiX adapts to changing part geometry, finishes, and materials—revealing subtle, specular, and depth-dependent defects that conventional setups miss. It supports true inline “fly-capture” on moving lines, minimizing downtime while elevating image quality for AI. Configured in minutes through an intuitive UI, OptiX accelerates deployment and changeovers, reduces maintenance complexity, and forms the imaging backbone of the FleX platform for high-mix, high-throughput production. OptiX Features: Software-Defined Imaging Graphical user interface setup of precise lighting patterns Sequential configuration for complex, multi-step lighting execution 2. Dynamic Patterns 2³² lighting patterns 75-degree incident angle range 3. High-speed imaging 50 lighting patterns/second 1.0 meter/second fly capture 100x brighter than conventional light Up to 1,800 parts/minute 4. 2.5D Technology Capture depth information with a single CMOS area scan camera 6 x 3D reconstructions per second at 5MP resolution Up to 500mm/s fly capture of moving parts 3µm z-repeatability for local defects at 500mm FOV More information please visit: | CorteX is the AI “brain” of FleX—a sample-efficient deep-learning and inference system built for production. It learns robust defect detectors from minimal real data (as few as three images, augmented via GenAI when needed) and runs them in real time on edge compute for inline decisions. With low-code OK/NG logic, centralized orchestration to sync models plant-wide in minutes, and native PLC/MES/FTP connectors for traceability, CorteX shortens ramp-up and keeps accuracy high. Continuous improvement is built in: human-in-the-loop reviews, SPC/OEE analytics, and versioned model management ensure performance stays aligned with quality and throughput targets. Main Features: Superhuman Accuracy Achieves superhuman accuracy w/ 9X lower escapes on high variance defects while not overkilling OK parts Sample Efficient A.I. Rapidly teach models with as few as 5 images per defect type Fast Decisions 30 inferences per second quickly pass a decision on OK/NG Adjustable Tolerances Defect classification with pixel precision and adjustable tolerances