Getting Ahead in Retail with Face Recognition

Companies are making use of increased connectivity, facial recognition software and the IoT to create personalized shopping experiences, writes Caroline Hayes, Senior Editor

Digital signage is evolving from a static display to a dynamic way to reach and engage with consumers. The combination of hardware, software and connectivity allows targeted promotions that can be renewed or replaced, according to market conditions.

AdMobilize is using connectivity which is everywhere to “make dumb assets smart.

One example of the combination of software and hardware is the use of facial and object recognition to ensure public safety and enhance consumer convenience. VIA Technologies is one company that is combining computer vision systems in kiosks in public areas, transportation, and retail. Its Smart Recognition Platform (Figure 1) is based on the SOM-9X20 system module, which is powered by the Qualcomm Snapdragon 820 quad-core processor. It supports object and facial recognition to detect, count, or track customers, staff and visitors and can deliver personalized advertising in a store according to the profile (age, gender) of faces detected. Immersive visual graphics are supplied by the Qualcomm Adreno 530 Graphics Processing Unit (GPU).

An integrated module with two antenna connectors provides wireless connectivity options, including Global Positioning System (GPS), Bluetooth 4.1 and Wi-Fi 802.11a/b/g/n/ac.

Artificial Intelligence (AI) provides facial and object recognition, which allows retail signage kiosks to deploy staff, visitors, or consumers to areas that are either lighter in traffic, for convenience, or to track people and goods to enhance security.

Facial Recognition
The module is compact, measuring 82 x 45mm (3.2 x 1.78 inches) yet includes 64-Gbyte eMMC flash memory and 4-Gbyte LPDDR4 SDRAM. There are Input/Output (I/O) and display expansion options with the MXM 3.0, 314-pin connector, USB 3.0, USB 2.0, High Definition Multimedia Interface (HDMI) 2.0, Secure Digital I/O (SDIO), PCIe, MIPI Camera Serial Interface (CSI), MIPI Display Serial Interface (DSI) and multi-function pins for Universal Asynchronous Receiver/Transmitter (UART), Inter-integrated Circuit (I²C), Serial Peripheral Interface (SPI) and General Purpose I/O (GPIO).

The Artificial Intelligence (AI) algorithm provides facial and object recognition, which allows retail signage kiosks, as well as building control systems and transportation systems, to control and monitor numbers and to deploy staff, visitors, or consumers to areas that are either lighter in traffic, for convenience, or to track people and goods to enhance security.

One of the critical aspects of the platform is that it is not only scalable but customizable. The SOMDB2 multi-I/O carrier board is also available. According to the company, this can accelerate system development. VIA offers technical support and design assistance to help developers create custom baseboards. Systems can be tailor-made to offer a range of services or technologies for individual applications.

Figure 1: VIA’s SOM-9X20 is at the heart of its facial recognition platform.

A Board Support Package (BSP) runs Android 7.1 and there is also the Smart Embedded Tool Kit (ETK), which has Application Program Interfaces (APIs) such as a watchdog timer, a real-time clock for auto power-on, and a sample app.

The company is also developing a BSP which supports Linux Kernel 3.18.44.

Display Options Count
The availability of display options is emerging as an important factor for Single Board Computers (SBCs) used in digital signage. AAEON offers the GENE-APL7 (Figure 2), compact motherboard with 10 USB ports and up to 12 COM ports. Dimensions are 5.75 x 4.0-inches (146 x 101.7mm).

This number of ports means there are no constraints, says the company, adding the board can be used for fintech as well as retail. The low power, fanless operation allows for a slim format for use in applications where space is limited.

There is one VGA and two LVDS connectors, which can be customer-configured to incorporate eDP technology. There are also MiniCard and mSATA expansion slots and a built-in speaker amplifier.

The board is powered by an Intel Pentium N4200 or Celeron N3350 processor with up to 8-Gbyte DDR3L memory.

Figure 2: The GENE-APL7 is versatile and slim and targets retail and fintech applications.

Voice Recognition
At the Digital Signage Expo 2018, AdMobilize introduced development boards to create Internet of Things (IoT) solutions using voice recognition commands. CEO Rodolfo Saccoman explains that the company is using connectivity which is everywhere to “make dumb assets smart.” The Matrix Creator and Matrix Voice development boards both offer voice recognition commands. The Voice version is a smaller form factor than the Creator, but both use open source Operating Systems (OS).

It can be integrated into systems in retail areas to respond to commands as simple as “Assistance, please” to send a message via Bluetooth to a sales assistant to let them know that a customer in a particular area of the store needs attention.

The Creator board (Figure 3) is based on Microchip’s Cortex-M3 ATSAM3S2 microcontroller and Xilinx’s Spartan 6 XC6SLX4 and has a microphone array consisting of eight Micro Electrical-Mechanical Systems (MEMS) MP34DB02 audio sensor digital microphones. Sensors include those for humidity and altitude as well as an UltraViolet (UV) light sensor and a Near-Field Communication (NFC) sensor for payments. It also has Bluetooth and Zigbee communications.

Figure 3: The Matrix Creator is one of two development boards connecting the IoT to digital signage.

The open source software development kit enables developers to code gesture triggers into any camera supported application. The company has also created IoT applications, such as Matrix Security, which uses computer vision to remotely monitor locations, and Matrix Intelligent Assistant (MIA), which uses voice and gesture commands.

Used with the Matrix OS and Matrix Command Line Interface (CLI), the board allows developers to build IoT apps that marry facial recognition with an action, for example to allow access to certain areas of the store.

The Matrix Voice board uses the same Xilinx FPGA, and also has an array of eight MEMS microphones. It has an optional ESP32 microcontroller, using a Tensilica Xtensa LX dual core microprocessor, for Wi-Fi 802.11b/g/n and Bluetooth 4.1 Low Energy wireless communications.

Here To Help
Taking a different approach, Panasonic mobilizes digital signage, in the form of the HOSPI, autonomous delivery robot (Figure 4). Already trialled in hospitals, the robot is in use for a trial period at the Narita International Airport and ANA Crowne Plaza, Japan. The Narita airport serves the greater Tokyo area.

“We have provided these robots to offer delivery services in hospitals [for drug and specimen deliveries],” says

Figure 4: HOSPI is the ultimate host—with drinks for visitors in its body and up-to-date information in three languages on its screen.

Ryosuke Murai, Manager Robotics Business Promotion Department, Panasonic Production Engineering. “We believe there are many environments outside of hospitals in which the HOSPI can be of service.” The robot is equipped with three displays, with messages shown in Japanese, Chinese, and English. It can navigate around the hotel lobby or airport to advise passengers and visitors when the next shuttle bus is leaving, other travel information, and departure announcements.

It navigates using pre-programmed map information, sensors, and collision-avoidance algorithms. It does not have to follow guide tapes to follow a route, nor equipment embedded into walls or ceilings as navigation beacons.

As the population ages and Japan faces a labor shortage, it is expected to use more robots for tasks such as delivery and hospitality services.

According to Grand View Research, the global digital signage market will nearly double from its 2016 figure of $16,044 million to $31,714 million by 2025 (Digital Signage Market Analysis By Type, By Component, By Technology, By Application, By Location, By Content Category, By Size, By Region, And Segment Forecasts, 2018 – 2025). Much of this growth is expected to be from new applications, increased advertising content, and the ability to manage and update information remotely. The use of AI and robotics will also contribute to the distribution of data and the targeting of information.

Caroline Hayes has been a journalist covering the electronics sector for more than 20 years. She has worked on several European titles, reporting on a variety of industries, including communications, broadcast and automotive.


Share and Enjoy:
  • Digg
  • Sphinn
  • Facebook
  • Mixx
  • Google
  • TwitThis


Extension Media websites place cookies on your device to give you the best user experience. By using our websites, you agree to placement of these cookies and to our Privacy Policy. Please click here to accept.