Retailers can use an edge AI appliance such as Lanner’s EAI-I130 instead of relying on cloud computing. The appliance connects to these cameras’ data feeds and provides AI and CV at the edge.  

With large data sets and intelligent models, retail can learn about consumers’ behaviors and preferences. Learning about their customers can, in turn, allow retailers to provide a seamless shopping experience and increase customer engagement.  

Requirements.  

Consumer demands, behaviors, and expectations are changing rapidly, especially due to the COVID-19 pandemics. According to McKinsey, for instance, only in the United States, about 75% of consumers have tried a new store, brand, or different way of shopping during the pandemic.  

To keep up with these changes (and maybe even influence them), companies must take a step forward by leveraging data-generated consumer insights.  

Retail businesses can use digital data-gathering techniques such as cameras and sensors to understand emerging demands and behaviors. Creating insights from these gathered data can help companies generate insights and enhance customer experiences.  

Computer Vision (CV), a field of AI, is one of those technologies that can help retail companies run analyses and that data and recognize images. Retailers can use CV for geofencing, recommendations-engine, and much more.  

However, Computer Vision can be challenging and complex to implement. It requires the following: 

  • AI (including CV) solutions are now being deployed in centralized cloud environments for easier management and scalable computing. Unfortunately, running Computer Vision in the cloud is heavily limited by the Internet connection. There is only so much data a retailer can send for real-time analytics. Retail would require a massive large bandwidth, ultra-low latencies, and high cloud costs. Using the cloud for vision limits real-time responses, risks data privacy, and makes off-grid applications impossible. 
  • Computer Vision needs lots of data, and to process such huge sets of visual data (while avoiding the cloud), users need the right type of computing hardware. Real-time data-intensive applications such as the retail industry require significant processing power.  

Solution 

AI implemented as close as possible or within the store (at the edge network) will transform retail. The imaging system collects large image sets in real-time through in-store video cameras, photo cameras, sensors, or 3D technology. Advanced computer vision algorithms analyze these images, extract information, detect anomalies, track objects, etc.  

An edge AI appliance is a high-performance computer connected to these cameras—deployed closest to the data source (at the store or in proximity to the store). The edge AI appliance allows on-device data processing rather than relying on cloud computing. The system could still send pre-processed data (non-sensitive, non-critical, and non-intensive data feeds) to the cloud for further analysis. Still, with edge capabilities, it could even provide AI and CV for off-grid applications.  

The Edge AI Appliance: EAI-I130 

Lanner's EAI-I130 is an industrial-grade AI inference system for the edge with NVIDIA® Jetson Xavier NX. The Jeston Xavier, NX Series Modules, is considered, by Nvidia, as the World's Smallest AI Supercomputers. They bring supercomputer performance to the edge in a small form factor System-On-Module.  

The edge AI appliance provides up to 21 TOPS— the maximum achievable AI performance from the SoC. With this kind of power, the EAI-I130 can run modern neural networks (including computing vision) and process data from multiple high-resolution cameras and sensors— all from the retail store.  

The EAI-I130 also highlights. 

  • Robust wireless support, 5G, and Wifi6. 
  • IP40 standard fanless design. 
  • -40°C To 70°C operating temperature range. 
  • 384-core NVIDIA Volta GPU With 48 Tensor Cores. 
  • 2x GigE PoE LANs With Support For IEEE 802.3 af/at PoE(+). 

Below is a diagram of how the retail computer vision analytics system works. The in-store cameras feed data to the edge AI appliance (EAI-I130), which in turn processes it and leaves it on the premises. The data (non-sensitive and non-critical) can be sent to the cloud for further analysis or remote monitoring. 

Applications.   

As of 2022, the majority of consumers are using multi-channel retailing experiences; they are both shopping online and going to traditional brick-and-mortar stores. Consumers are also looking for seamless and unique buying experiences, which is critical for the sustained growth of any business.  

The edge AI Computer Vision solution can provide multiple applications for both brick and mortar retail and ecommerce, and improve customer’s experiences.  

The core benefits of the EAI-I130-based retail vision analytic solution is an improved customer experience and engagement, enhanced operational agility, and seamless omnichannel management. With Computer Vision and image analyses at the edge, retailer companies will learn to foresee what consumers will do next.  

Applications and Benefits.  

  • Image Recognition in Retail. Build image recognition systems to help establish facial recognition systems to identify employees and grant them access to restricted areas like stockrooms, and offices. A face recognition system in retail could also improve customer experience, by allowing touchless payment systems.  
  • Inventory and Shelf Space Management. Computer Vision can help inspect store shelves in order to determine whether items are placed efficiency, their pricing is correct, or detect product stockouts and shrinkages. This could help keep a more efficient and up to date inventory and improve in-store supply.  
  • Crowd Analysis. Real-time Computer Vision can be used by retailers, to count customers and analyze their behaviors when moving in multitudes. An example of crowd analysis is ensuring a store keeps up with health regulations such as social distancing or maximum occupancy. Another example would be to keep track of the customer’s journey throughout the store.  
  • Consumer behavior analysis. Track shoppers on the store and generate reports of their activities, including passer-by traffic, impressions, dweller events, and interactions. In addition, create a retail heat map that helps understand all store interactions.  

Next Steps.  

For more information on the edge AI appliance or retail computer vision solution please contact Lanner’s sales representative. 

Featured Product 

EAI-I130 


EAI-I130

Industrial Grade AI Inference System For 5G Edge With NVIDIA® Jetson NX

CPU 6-core NVIDIA Carmel ARM®v8.2 64-bit CPU 2MB L2 + 4MB L3
Chipset N/A

Read more