Quividi, the pioneer and World’s number one audience and campaign intelligence platform for DOOH and Retail Media, has integrated Plumerai’s people detection AI into its latest Anonymous Video Analytics (AVA) solution.
With this partnership, Quividi now offers real-time, long-distance impression counting in its Audience Measurement Platform on entry-level CPUs, which commonly power smart machines or retail shelf screens.
Vending machines, coolers, kiosks, digital merchandising displays and retail media screens now have access to Quividi’s high-fidelity audience impressions to monetize their screen networks directly or programmatically while being able to leverage Quividi’s real-time shopper engagement data to drive sales up.
Over the years, Quividi has developed a hybrid approach to computer vision with the in-house development of the core algorithms powering its audience measurement platform, as well as the overall architecture, while integrating and optimizing sophisticated third-party inference models. This modularity makes the Quividi platform the most versatile and accurate solution to provide audience insights for small to large displays, in-store, indoor and outdoor.
With the integration of Plumerai’s solution, Quividi carries on expanding the range of its platform capabilities. Plumerai’s AI models are tiny and fast and have been designed to run seamlessly on Arm Cortex-A processors. Trained with 30 million images, their AI detects human bodies with high accuracy in any retail environment.
“Quividi prides itself on integrating with the most advanced AI companies to provide the best AVA platform for DOOH and Retail Media,” says Olivier Duizabo, President of Quividi. “With Plumerai, Quividi complements its offering to smart machines and retail media companies, opening new avenues for audience measurement and monetization”.
“We’re honored that our technology has been selected by the market leader for audience measurement solutions,” says Roeland Nusselder, Plumerai’s CEO. “Our deep-learning-based people detection runs on tiny Arm CPUs, instead of costly and power-hungry hardware. This enables Quividi to bring its solution to many more machines and retail locations. We’re looking forward to working together with Quividi for many years to come.”
Quividi is organizing a special webinar on audience measurement for smart machine operators and retail media & merchandising networks on October 25th, 2022. You can register for free here.
Plumerai develops production-worthy embedded AI through a relentless focus on the full AI stack, from collecting and curating data, to training algorithms, model architectures, inference engines, and hardware optimizations. Their people detection AI consistently proves to be the most accurate in any environment, while consuming minimal resources. With a tiny memory footprint of 1MB, it runs efficiently on nearly any hardware architecture, reaching 55 frames/s on a single-core Arm Cortex-A CPU. Their inference engine for Arm Cortex-M is the fastest and smallest in the world. Their Ikva AI accelerator fits in the smallest of FPGAs. Plumerai has offices in London, Amsterdam and Warsaw.