State-of-the-art Edge AI from Infineon with DEEPCRAFT™ software and PSOC™ Edge

Recognize and track faces with DEEPCRAFT™ Face ID. 

• Unparalleled Accuracy: State-of-the-art cascaded neural network architecture delivers >96% accuracy with extremely low false positives (1 in 1 million)
• Multi-Face Detection: Recognize and track multiple faces simultaneously
• Scalable Enrollment: Designed to handle large numbers of face enrollments, with testing up to 1500 unique faces
• Real-Time Performance: Low latency face recognition at speeds over 25 FPS
• Efficient Power Consumption: Ultra-low power face identification, accelerated by built-in NPU
• Enhanced Privacy: Ensures encoded user representation for secure face recognition
• Streamlined Enrollment: Tooling for fast and accurate face enrollment, simplifying the onboarding process

This demo showcases our robot for outdoor surface data collection. Specifically, we demonstrate our DEEPCRAFT™ AI model for detecting if a surface has grass or no-grass using data from a BGT60TR13C radar. The model runs in real-time on PSOC™ 6.

• Multi-sensor: Collect data from different sensors (radar, RGB cameras, ToF cameras, IMU, CAPSENSE™, etc.) using ROS system 
• High accuracy: Edge AI model running on a robot/PSOC™ 6 MCU for grass detection using BGT60TR13C radar data
• Built for the edge: works with PSOC™ 6, PSOC™ Edge or similar devices that can handle BGT60TR13C radar data
• Quick detection on PSOC™ 6: model outputs up to 10 predictions per second
• Differentiates grass from other surfaces: gravel, sand, asphalt, earth, stone, cement, metal, manhole, grid, etc.
• Autonomous: robot reliably stays within boundaries of lawn 
• Robust: against ambient light (direct sunlight, no-light)
• Obstacle and cliff detection could also be included