STMicroelectronics Debuts Ultra-Low-Power VD65G4 and VD55G4 Sensors

STMicroelectronics introduced two ultralow-power global-shutter CMOS image sensors, the VD65G4 (RGB) and VD55G4 (monochrome), aimed at battery-powered edge vision for wearables, AR/VR, smart home and medical devices, per ST product pages and the company press release reported by Business Insider. The sensors use a compact 1/9-inch optical format and a 2.16 um pixel pitch and deliver 0.56 MP (804 x 704) resolution, up to 184 fps at full resolution, and outputs in RAW8/RAW10, according to ST and CNX-Software. Power consumption is reported at 35 mW typical @ 60 fps, 1-2 mW in auto wake-up mode and 0.8 mW in standby, per ST documentation and press coverage. "Always-on vision is becoming essential... With VD55G4 and VD65G4, we are bringing this capability to smaller, lighter products," said Alexandre Balmefrezol, Executive VP of ST's Imaging Sub-Group, in the press release. Editorial analysis: These sensors target event-driven, always-on use cases where sensor-level preprocessing and milliwatt sleep modes materially affect system battery life.
What happened
STMicroelectronics introduced two new ultralow-power global-shutter CMOS image sensors, the VD65G4 (RGB Bayer) and VD55G4 (monochrome), as detailed on the ST product pages and in the company's press release reported by Business Insider. Per ST's product documentation and CNX-Software coverage, both sensors provide 0.56 MP (804 x 704) resolution, a 2.16 um global-shutter pixel, a 1/9-inch optical format, and raw outputs in RAW8 and RAW10. Frame-rate capabilities are listed as up to 184 fps at full resolution, 271 fps at VGA and 480 fps at QVGA, according to CNX-Software and ST's specification sheets. ST and press coverage report typical power use of 35 mW at 60 fps, with an "auto wake-up" event-driven mode drawing 1-2 mW and a standby mode at 0.8 mW.
Technical details
Per ST's product pages and the ST blog, the sensors combine back-side illuminated pixels, CDTI and 3D stacking to achieve a bare-die footprint of 2.73 x 2.16 mm while supporting MIPI CSI-2 (1-lane), I3C and SPI control/data options. On-sensor features documented by ST include hardware autoexposure, automatic dark calibration, defective-pixel correction, background removal (image-difference mode), programmable frame statistics and context management for up to four contexts. ST describes a low-frame-rate autoexposure and auto-white-balance mode that precomputes exposure and color parameters on-sensor to reduce host CPU/NPU load and end-to-end latency, per the product page.
Industry context
Editorial analysis: Devices constrained by battery and size, such as smart glasses, XR headsets and wearables, increasingly rely on sensor-level intelligence and ultralow-power wake strategies to enable continuous situational awareness without heavy host compute. Observers following the sensor-to-host data-rate problem note that shifting preprocessing into the sensor and supporting higher-bandwidth but low-pin interfaces can reduce MCU wake-ups and system-level power. ST's documentation and demo focus on event-driven wake-up and richer statistical outputs, which match broader trends toward sensor-side feature extraction for always-on edge vision.
Implications for practitioners
Editorial analysis: For embedded vision engineers, these sensors offer a compact option where milliwatt-scale sleep and on-sensor preprocessing reduce the need for frequent MCU or NPU wake cycles. The availability of 1-lane MIPI CSI-2 plus higher-speed I3C or SPI control paths simplifies connection to low-power microcontroller platforms, which can lower BOM and software complexity for simple presence, glance or face-detection tasks. Integrators should treat on-sensor statistics and background-removal modes as real avenues to reduce host compute, while validating end-to-end latency and detection false-positive rates in their specific environment.
What to watch
Editorial analysis: Watch for early-adopter designs and reference demos that quantify system-level battery gains using the auto wake-up mode. Also monitor third-party module availability, lens and mechanical integration for the 1/9-inch format, and any SDK or example firmware from ST that demonstrates interfacing the sensor statistics into lightweight inference pipelines. Finally, evaluate the sensors in low-light and NIR scenarios for the VD55G4 monochrome variant, as real-world performance will determine suitability for presence and glance-detection use cases.
"Always-on vision is becoming essential for the next generation of personal electronics, from smart glasses and AR/VR headsets to intelligent home appliances and medical devices. With VD55G4 and VD65G4, we are bringing this capability to smaller, lighter products that must run for a long time on a tiny battery," said Alexandre Balmefrezol, Executive Vice President and General Manager of the Imaging Sub-Group at STMicroelectronics, in the company press release reported by Business Insider.
Scoring Rationale
This is a notable product launch for embedded and edge-vision practitioners because the sensors materially lower always-on power budgets and integrate sensor-side preprocessing. The story is product-level, not a paradigm shift, so it scores as a meaningful hardware advance for device integrators.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems


