Automatic Licence Plate Recognition (ALPR) is becoming increasingly common in and around modern cities. Pay-by-plate tolling, parking garages, and even speed cameras all use licence plate numbers to collect fees. Many of these systems rely on cloud computing and/or humans in the loop to accurately identify plate IDs. This is an expensive and energy intensive operation that relies on transmitting large numbers of high-resolution images over the internet.
In this whitepaper, we explore how XMOS’ accurate, low power and cost-efficient Automatic Licence Plate Recognition (ALPR) solution enables service providers to take parking management from a complex resource-intensive system, to simple on-device AI.
Combining the fast, flexible and economical xcore®.ai and light neural network model with plate detection and optical character recognition, the new generation ALPR moves away from heavy computing on cloud and high-cost cameras, offering the low BOM ecological turnkey solution.
