Facial recognition system developed for smartphones

Professor Hoi-Jun Yoo of the Department of Electrical Engineering, along with his research team and collaboration with start-up company, UX Factory Co, has developed a semiconductor chip, CNNP (CNN Processor), which runs AI algorithms with ultra-low power, and K-Eye, a face recognition system using CNNP.

Consisting of two different formats, the K-Eye series is available as a wearable type and a dongle type. The wearable type device can be used with a smartphone via Bluetooth, and it can operate for more than 24 hours with its internal battery. By conveniently hanging the K-Eye around their necks users can check information about people by using their smartphone or smart watch, which connects K-Eye and allows users to access a database via their smart devices. A smartphone with K-EyeQ, the dongle type device, can recognise and share information about users at any time. 

It works by recognising an authorised user looking at the screen, which then automatically turns the smartphone on, without a fingerprint, passcode or iris authentication. The smartphone cannot be tricked by the user’s photograph, as it can distinguish whether an input face is coming from a saved photograph versus a real person.

Other distinct features are carried out by the K-Eye series. Detecting a face at first and then recognising it is one, and it is possible to maintain ‘Always-on’ status with low power consumption of less than 1mW. The research team devised two key technologies to complete this: an image sensor with ‘Always-on’ face detection and the CNNP face recognition chip. 

The ‘Always-on’ image sensor, the first key technology, is able to determine if there is a face in its camera range. Then, it can capture frames and set the device to operate only when a face exists, reducing the standby power significantly. Additionally the face detection sensor combines analogue and digital processing to reduce power consumption.

Using this approach, the analogue processor, combined with the CMOS Image Sensor array, distinguishes the background area from the area likely to include a face, and the digital processor then detects the face only in the selected area. Therefore, it becomes effective in terms of frame capture, face detection processing, and memory usage. 

Following this the second key technology, CNNP, is able to achieve incredibly low power consumption, by optimising a convolutional neural network (CNN) in the areas of circuitry, architecture, and algorithms. Specially designed to enable data to be read in a vertical direction as well as in a horizontal direction, the on-chip memory integrated in CNNP also has immense computational power with 1024 multipliers and accumulators operating in parallel and is capable of directly transferring the temporal results to each other without accessing to the external memory or on-chip communication network. Additionally, convolution calculations with a two-dimensional filter in the CNN algorithm are approximated into two sequential calculations of one-dimensional filters to achieve higher speeds and lower power consumption. 

CNNP achieved 97% high accuracy but consumed only 1/5000 power of the GPU thanks to these new technologies. Face recognition can be performed with only 0.62mW of power consumption, and the chip can show higher performance than the GPU by using more power. 

Developed by Kyeongryeol Bong, a PhD student under Professor Yoo, these chips were presented at the International Solid-State Circuit Conference (ISSCC) held in San Francisco earlier this year.

CNNP, which has the lowest reported power consumption in the world, has achieved a huge amount of attention, which has led to the development of the present K-Eye series for face recognition. 

Professor Yoo commented: “AI - processors will lead the era of the Fourth Industrial Revolution. With the development of this AI chip, we expect Korea to take the lead in global AI technology.” 

You can find more about the K-Eye technology here.