XMOS releases low-cost, low-power ALPR reference design for smart parking

Get your number: XMOS releases low-cost, low-power ALPR reference design for smart parking

Designed to help move solutions from high-cost, high-power hardware to streamlined, efficient machine learning

Bristol, UK, XX March 2022 — XMOS, the UK’s leading chip company, today announced the launch of its Automatic Number Plate Recognition (ALPR) reference solution , designed to move ALPR in car parking from complex, resource-intensive hardware to simple on-device AI.

Developed in partnership with computing specialist Cloudtop, the reference design can read slow-moving license plates at a distance of 3-5 meters with high accuracy. Thanks to the capabilities of XMOS’ xcore.ai chip, Cloudtop’s machine learning models – originally designed to process high-resolution video frames – have been seamlessly adapted to work in low-power, low-cost scenarios without sacrificing accuracy.

Parking lots using ALPR have traditionally integrated hardware that far exceeds the specifications required for slow-moving, close-range license plate recognition. High-resolution cameras run on complex machine learning models that rely on cloud connectivity for image processing, making ALPR prohibitively expensive to implement in many cases.

XMOS’ reference design provides the required power and device intelligence, significantly reducing power consumption and bill of materials (BOM) compared to standard ALPR solutions. While eliminating the need for high-cost hardware and nearly eliminating the need for cloud connectivity, this device becomes a realistic component of the entire smart city ALPR infrastructure.

“For smart parking, cloud connectivity and massive processing power is simply overkill,” commented Aneet Chopra, Vice President of Products, Marketing and Business Development at XMOS. “It makes the ALPR network much more expensive than it needs to be, makes it more complex to maintain, and is fraught with the privacy issues inherent in cloud computing.

“The reference design we developed eliminates those issues by simplifying the process. If you can provide the required intelligence and functionality on the device, you can avoid sending all the raw data to the cloud or to overly expensive or powerful hardware. In the long run, this can only help us advance ALPR.”

“Simplicity and affordability are two priorities in the ALPR field that not only drive sales but also encourage innovation,” commented Professor Zhang, co-founder of Cloudtop. “Making devices cheaper, simpler, and more reliable is important for smart cities, and scaling down machine learning models so they can run on chips like xcore.ai that can be mass-produced gives developers both funding and design flexibility to experiment.”

XMOS and Cloudtop will be showcasing the solution at the tinyML Summit in San Francisco from March 28-30, and invite all attendees to visit their booth and poster presentation.