1. The Metavision® Event Pipeline
Traditional cameras capture redundant frames. Prophesee's Metavision sensor architecture captures only *changes* in light at the pixel level, creating a sparse, high-speed temporal event stream. Click to explore the asynchronous flow.
Pixel Trigger
Event Stream
Algorithm Engine
Action Output
Asynchronous Pixel Trigger
Each pixel in the Metavision sensor acts independently. It only sends data when it detects a relative change in light intensity (temporal contrast).
Input Data
- Incident Photons
- Temporal Contrast Thresholds
Output Data
- Discrete Event Trigger
- Pixel Address (X, Y)
2. Metavision vs. Frame-Based Workloads
Prophesee sensors operate with **microsecond precision**, effectively delivering temporal resolution equivalent to >10,000 fps while consuming 10x–100x less power. The data is sparse, meaning compute is only expended where motion occurs.
🔍 Key Performance Metric
Metavision sensors achieve >120dB dynamic range, allowing clear vision in extreme lighting (e.g., staring directly into the sun while seeing details in shadows).
3. Foundational Research
The Metavision paradigm is supported by over a decade of neuromorphic engineering research and collaboration with industrial partners like Sony.
Event-based Vision: A Survey
Gallego et al. (IEEE TPAMI)
The definitive academic survey of event-based sensing, processing, and application paradigms.
Sony/Prophesee Stacked Sensor
ISSCC 2020 Whitepaper
Technical details on the first industrial-grade HD event-based vision sensor (IMX636).
Metavision Intelligence SDK
Prophesee Documentation
Proprietary algorithms for optical flow, tracking, and vibration analysis in sparse event spaces.
Strategic Partnership Opportunities
Prophesee's technology provides a massive competitive advantage for high-speed industrial, mobile, and safety-critical applications.
Industrial Automation
High-speed counting (>1,000 obj/sec) and predictive maintenance through vibration analysis.
Automotive ADAS
Ultra-fast obstacle detection and driver monitoring systems (DMS) with microsecond latency.
Mobile & AR/VR
Low-power eye tracking and hand gesture recognition for next-gen spatial computing devices.