Industrial edge cameras require low-power solutions.
Maxim and Xiliant Develop Facial Recognition NN MCU
Power consumption is 1/250th that of existing embedded solutions.


Battery-powered AI systems that require facial recognition, such as industrial smart security cameras, home cameras, and retail cameras, require low-power solutions that operate for as long as possible.

Maxim Integrated Korea announced on the 28th that its 'MAX78000' ultra-low-power NN MCU has been enabled to recognize faces in edge video and images using AI specialist Xailient's Detectum neural network (NN) technology.

Maxim's MAX78000 MCU, paired with the Xyliant Detectum NN, not only supports standalone applications but also optimizes low-power sleep (listening) mode for hybrid edge/cloud applications that operate complex systems during facial detection, improving power efficiency and extending overall battery life. This extends the operating time of coin-cell-based hybrid edge/cloud applications by up to several years.

Faster AI inference speeds can improve accuracy through real-time or average multi-inference. Xyliant's NN consumes 1/250th the power of existing embedded solutions and recognizes faces at a processing speed of 12ms per inference. Additionally, the system incorporates focus, zoom, and wake word recognition technologies, enabling facial recognition in video and images at speeds comparable to or 76 times faster than existing software solutions. Furthermore, its flexible network allows for expansion into diverse applications, including livestock operation management and monitoring, parking recognition, and livestock counting.

“Thanks to Xiliant’s Detectum NN, the Maxim MAX78000 MCU can perform both classification and localization, enabling it to not only recognize faces in images or videos but also determine where those faces are located in the field of view,” said Robert Muchsel, senior analyst at Maxim Integrated. “It can also implement things like the number and presence of people, vehicles, and objects, obstacle recognition and path mapping, and customer flow heat maps.”

“AI is the second-largest carbon-emitting industry,” said Shivy Yohanandan, CTO and developer of Xyliant’s Detectum NN technology. “Replacing 14 legacy internet protocol cameras using traditional cloud AI with edge-based cameras powered by Maxim’s MAX78000 and Xyliant’s NN has the same carbon footprint as removing one gasoline vehicle from the road.”