Processing video analytics in retail –edge or cloud?
In the previous articles we explored how retail video analytics could address various issues facing retailers. More and more, these analytics are moving to the edge due to various advantages. This article discusses the latest in edge AI vs. cloud AI in retail surveillance.
Offline retailers face a variety of challenges nowadays. Among them are loss and shrinkage due to criminal activities or mismanagement. Meanwhile, storeowners are constantly seeking to improve the shopping experience, amid fierce competition from offline and online counterparts.
In this regard, video analytics in retail, a lot of them AI-based, can play a key role. “In practical terms, the metadata from AI is speeding up investigations, cutting down the time it takes to find footage of interest from hours to minutes. It is also very helpful in tracking down missing children in crowded stores or
malls. And, it can be an important tool in helping to reduce human error, which is a major contributor to shrinkage, through its ability to rapidly retrieve and review events,” said Jason Burrows, Regional Sales Director for Western U.S. at IDIS.
He added: “Increasingly more stores – of all sizes – want to leverage analytics-driven metrics of activity in-store to increase sales and build customer loyalty. Most are leaning toward an essential group of analytics
functions that give then in-store and customer intelligence that include people counting, queue monitoring, heat mapping, and occupancy monitoring so they can gain real-time and historic activity data from specified locations.”
Edge AI vs. cloud AI
With video retail analytics becoming increasingly popular, the question then becomes whether video analytics in retail should be hosted on the edge or in the cloud. Indeed, cloud computing for retail offers certain advantages. Yet more and more, users are also realizing the benefits of running retail analytics at the edge, inside cameras or edge appliances.
There are at least two advantages of the edge retail concept. “Analytics on the edge of the network – within the camera itself – delivers the best-quality solution. These processes which take place within the camera mean that only the valuable data needs to be transferred to the operator. Conversely, analytics taking place on the server requires that all of the data from the camera is transferred to the data center for analysis, and with that comes a much greater need for costly bandwidth,” said Atul Rajput, Director of
Channel Partners and End Customers for EMEA at Axis Communications.
“Secondly, analyzing video within the camera and as close as possible to its capture means that the images being reviewed are of the highest possible quality: there is no degradation that can come through compressing images prior to transfer, which is often done to reduce the former issue, ironically,” he said.
More powerful IP cameras
At the same time, IP cameras now have more and more processing power to run these analytics. “Following Moore’s law, network cameras have evolved to offer powerful capabilities, including edge-based deep learning processing reducing the requirement for separate, dedicated edge appliance
units. With network cameras trending towards using open software frameworks and industry standard APIs, including deep learning toolchains, it has become more attractive for new developers focused on AI in server and cloud environments, to deploy at the edge in a network camera device,” Rajput said.
This, then, may prompt users to upgrade retail security systems to adopt store surveillance cameras embedded with retail video analytics.
“Processors inside edge AI cameras are now powerful enough to run analytics locally, while still encoding and streaming without the need or cost to upgrade software. This capability is now becoming more widely available with 5MP domes and bullets. Once retailers are ready to upgrade to edge cameras, they will deliver faster insights and better security, while overcoming bandwidth constraints and storage burdens,” Burrows said.
AI edge boxes
For those that are more budget constrained and want to keep their existing retail store security cameras, AI edge boxes are a viable option.
“Extending the life cycle of existing surveillance systems is likely to become more important in the current economic climate if upgrades are put on hold. And if store performance is a challenge, rather than upgrading to edge cameras, an AI box appliance is a lower cost option to support adapting store layouts and improving merchandising and staff utilization to boost profitability. They can also provide both store managers and head offices the insights\ they need to make improvements that can have a significant impact on sales performance, rather than simply closing low performance stores, many of which are often the life blood of smaller towns and communities,” Burrows said.
He added: “AI boxes are ideal for retailers with small, often geographically spread branches, such as convenience stores, pharmacies, fashion chains, and beauty stores that want to benefit from essential analytics features without the cost, waste, and complexity of rip-and-replacing cameras or investing in expensive software.”
In the end, it needs to be noted that while edge computing has gained traction in retail, cloud and edge both have their merits. Retailers should choose an architecture that suits their unique requirements. In fact, discussions on edge AI vs. cloud AI in retail will not be so clear-cut anymore.
“Retail analytics in the future will not be question of choosing to deploy on the edge or in the cloud, but a distributed application that run across both platforms. This will allow retailers to better leverage all of the processing power within their system, thereby reducing costs and improving efficiency,” Rajput said.
Disclaimer – This post has only been shared for an educational and knowledge-sharing purpose related to Technologies. Information was obtained from the source above source. All rights and credits are reserved for the respective owner(s).
Keep learning 📚 and keep growing 📈