Prophesee Event-Based Vision Sensors & Mantara

Prophesee is a Paris-based deeptech company that has spent more than ten years developing event-based vision, a sensing approach inspired by the human retina. Instead of capturing full frames at fixed intervals, each pixel reacts independently to changes in light, which removes motion blur and cuts the data and power needed for fast-moving scenes.

Its portfolio combines event-based sensors developed with partners such as Sony, camera systems (Mantara, Mantara Lite, Nexara), the Embra exploration kit and the Hearth software platform with embedded AI models, which succeeds the Metavision SDK. Industrial uses include high-speed counting, vibration monitoring, particle sizing and defect inspection, while Mantara targets drone detection for airports, energy sites, ports and defence.

• Event-based sensing: each pixel reports changes independently with microsecond timing and no exposure time.
• High-speed counting: counts pills, parts or particles at over 1,000 objects per second.
• Vibration monitoring: contactless, multi-point vibration measurement for predictive maintenance.
• Inspection and gauging: surface defects, object gauging and particle tracking up to 200 kHz.
• Mantara system: integrated event-based drone detection and tracking that emits no signal.
• Hearth platform: software with embedded AI models, over-the-air updates and sensor fusion.

Pharmaceutical and food plants in Vietnam can count tablets or granules on high-speed lines. Electronics and mechanical factories can monitor motor and spindle vibration without attaching sensors. Airports, ports and power plants can evaluate Mantara as a passive layer for drone detection.

Microsecond pixel response and more than 120 dB dynamic range keep fast objects visible in low light and backlit scenes. Industrial counting reaches over 1,000 objects per second with more than 99.5% accuracy, and vibration can be measured remotely from 1 Hz into the kHz range without contact sensors. Sensing power below 10 mW and sparse data reduce compute load. Event data contains no identifiable images, which eases privacy compliance.

Factories are adding machine vision for faster lines and predictive maintenance, but frame-based cameras struggle with motion blur, lighting and data volume. Neuromorphic and event-based sensing is moving from research into industrial, robotics and security products, and the spread of small drones is raising demand for passive detection at critical sites. Manufacturing growth in Vietnam and Southeast Asia expands the market for high-speed inspection.

Headquartered at 75 ter rue de Charonne, Paris, with offices in Grenoble, Shanghai and Tokyo. Holds around a hundred patents. Raised 20 million EUR in June 2026 led by Critical Path Ventures with existing shareholders. Ecosystem partners shown on its site include Sony, Bosch, iCatch, FRAMOS, Century Arks, LUCID Vision Labs and IDS; signed an MoU with Exosens.