CEA-Leti develops autonomous face biometrics imager for mobile devices and cars
France-based research institute Leti has unveiled an autonomous imager technology capable of activating mobile devices and small appliances via facial recognition, which it says is the first of its kind.
The new µWAI (micro-WAY) biometric device is as small as a coin, and features a readout and processing architecture with an optimized artificial intelligence (AI)-powered algorithmic pipeline.
This means the recognition results from a sequence of elementary algorithms are used to provide ultra-low-power wake-up modes and compact silicon implementation.
According to CEA, a firm specializing in miniaturization technologies and Leti’s parent company, µWAI is the first-ever created sensor offering auto-exposure for all lighting conditions and 88dB dynamic range.
In addition, µWAI also features motion detection and biometric feature extraction capabilities for event-based functioning, as well as highly precise (95 percent reported accuracy) AI-based object recognition carried out at the network edge.
“The recognition engine is optimized to recognize faces when movement is detected,” explained Antoine Dupret, CEA-Leti’s industrial partnership manager.
The device’s advanced features not only help manufacturers to keep costs down but also enable highly reliable decision-taking within a very small pJ/pixel/frame. For context, the term refers to the measure of energy spent by each pixel for every single image within a frame of images.
CEA-Leti’s autonomous imager device µWAI
A typical implementation of the technology by a low-power camera plus processor set today roughly requires 10,000 times more energy than µWAI, according to the announcement.
The new biometric imager is powered by an always-on CR1025 battery, which should last around five years, and features a 3-6µW power rating, which is used in most internet of things (IoT) applications today.
Moving forward, Dupret said CEA-Leti will utilize its partnership with STMicroelectronics to explore further applications of the technology.
“CEA-Leti’s team is working hand-in-hand with STMicroelectronics to develop specific smart-imager products as we consider extending the technology to other use cases,” Dupret said. “We target adapting the recognition engine as IP embedded in various cameras and optimizing the performance of the imager to the requirements of our partner’s customers.”
CEA-Leti will officially present µWAI and its applications at the company’s Innovation Days event, coming up on June 22 and 23.