OmniVision Launches AI Coprocessor for Low-Power Parking Surveillance

OmniVision has released the OAX7700, an always-on visual AI coprocessor designed for automotive applications. The chip integrates a Neural Processing Unit (NPU) capable of object and human detection, as well as distance measurement. Its primary application is to upgrade existing automotive camera systems into low-power parking surveillance solutions.

The OAX7700's key advantage lies in its drastically reduced power consumption. OmniVision states that parking surveillance systems built with this processor require only 3% of the power needed by comparable systems currently on the market. This allows for continuous monitoring of the vehicle's surroundings with minimal energy draw.

In operation, the camera system maintains a constant, low-power monitoring state. Upon detecting a trigger event, such as motion or an object, it instantly switches to a higher performance mode. The camera captures a short 5-10 second pre-recorded video clip, which is then sent to the electronic control unit for further analysis and potential alarm activation.

The OAX7700 is housed in a 5.5 x 3.5 mm BGA package and includes stacked PSRAM and flash memory. Samples are currently available, with mass production slated to begin in the first quarter of 2027. This technology aims to enhance vehicle security during parking while significantly lowering energy costs.