Event cameras
Dynamic-vision-sensor cameras for high-speed, high-dynamic-range, low-power perception — available now.
← All companies
DVSense 动微视觉
From the DVS inventor's lab — event-vision sensor-compute chips
Event-based vision company rooted in the ETH Zurich / University of Zurich Institute of Neuroinformatics lineage; ships event cameras, RGB-plus-event fusion cameras, SDKs and application solutions today, with a 1080p sensor-compute chip planned for MPW in 2027 and volume production in 2028.
DVSense builds event-based ('bionic eye') vision sensors and is moving the computation into the eye itself. Its founder studied under Professor Tobi Delbruck, inventor of the dynamic vision sensor, and co-published DVS optical-flow research with him; the team covers event-stream algorithms, AI acceleration, perception algorithms and chip architecture. Event vision sees where conventional cameras are blind — high-speed motion, strong backlight, extreme light changes, always-on at low power — and DVSense already has solutions for high-speed robot manipulation and autonomous driving, with close to ten million renminbi of revenue in 2025, its first commercial year.
01Products and solutions
RGB + Event fusion cameras
Cameras combining a conventional frame sensor with an event sensor for applications that need both texture and microsecond motion information.
SDK and algorithm suite
Event-stream processing, optical flow, detection and perception algorithms with AI acceleration, for integration into robots and vehicles.
Application solutions
Ready-made solutions for high-speed robot manipulation, autonomous driving and other Physical AI scenarios.
Sensor-compute event-vision chip (in development)
A 1080p / ~2-megapixel event sensing array with 30–35 on-chip high-level operators, moving acquisition, processing and detection onto one chip; MPW in 2027, volume production planned for 2028.
02Key strengths
- Source team of DVS technology: founder Liu Min is a doctoral student of Tobi Delbruck, the inventor of the dynamic vision sensor, and co-published DVS optical-flow research with him.
- From 'bionic eye' to 'computing inside the eye': a 1080p sensor-compute chip that defines front-end perception computing for Physical AI.
- Solves the physical blind spots of conventional cameras — high-speed motion, strong backlight, extreme brightness changes, low-power always-on — with concrete solutions for high-speed robot operation and autonomous driving.
- Demand validated ahead of the chip: cameras, algorithms and solutions generated close to ten million renminbi in 2025 — 'validate the scenario first, then put it in silicon'.