3D Behavioral Map
Grazper delivers a complete embedded solution for real-time, wide-area, 3D analytics using deep learning. The system is video based, centers around edge computing and allows automated, highly-detailed person and object tracking.
Features
- Multiple cameras are combined on site with embedded hardware to obtain an advanced real-time model of the surroundings
- Highly-accurate people and object identification and 6 Degrees of Freedom tracking in a unified 3D world model
- Privacy-by-design, GDPR-compliant, anonymous tracking and re-identification
- Embedded; inference on the edge with plug-and-play setup
- IoT-ready; low bandwidth, latency & power
- Interoperability; interfaces with existing systems and business logic through an easy API

Technologies
At the core of our technology sits a full-stack, deep learning-based AI engine for embedded systems.
- Efficient, unsupervised learning for adaptive feature and object matching
- Minimal networks through binarization and weight compression
- Video annotation tool suite with semi-supervised learning
High-speed inference on multiple embedded platforms: CPU, GPU, FPGA.