新世界I&C发布AI视觉自助收银方案Spharos ScanFree

Shinsegae I&C (035510) unveiled "Spharos ScanFree" on the 28th, a self-checkout solution that uses artificial intelligence (AI) vision to instantly recognize products simply placed on the counter, without scanning each barcode individually.
The product recognizes multiple items simultaneously within 0.4 seconds at 99.9% accuracy. This is faster than the previous solution, which recognized products at 99.5% accuracy in one second. According to Shinsegae I&C, customer waiting time is reduced by up to ninefold compared with standard self-checkouts.
The range of recognizable product units has also been expanded to some 5,000 SKUs. It can recognize a variety of items without barcodes, including manufactured goods as well as fresh food such as fruits and vegetables, and even bread, which varies in shape even for the same product.
A function that asks customers for confirmation in ambiguous cases has also been added. When the AI has difficulty making a judgment because there is little difference in shape or color, it presents candidates for the customer to choose from rather than reaching an arbitrary answer. The AI learns from customer feedback in real time, so recognition accuracy improves with use.
"Spharos ScanFree" is currently deployed at a smart store within SK hynix's Icheon campus. Discussions on adoption are underway with a range of retail stores, including bakeries, ready-to-eat food specialty shops, and household goods stores.
"Spharos ScanCare," introduced earlier in 2024, is also delivering results. Spharos ScanCare is a solution that uses AI vision to analyze self-checkout usage in real time and guide proper use. It detects situations where items are not scanned properly due to customer mistakes or barcode errors, guides customers to rescan, and sends notifications to operators when necessary. At one retail store that adopted it, the frequency of customers calling staff dropped to one-fifth compared with standard self-checkouts, and losses from missed scans were cut by up to tens of billions of won annually.
Shinsegae I&C's AI vision-based solutions are designed with an edge AI architecture that processes data such as images and video within the store, without a separate server or external graphics processing unit (GPU). Because data is processed only within the device, it is advantageous for information protection, and it can be adopted without changing existing store systems. The adoption cost is about one-quarter that of other companies' solutions. Based on the performance verified in the domestic market, Shinsegae I&C plans to target Asian markets including Japan.
"Having completed the advancement of our solution based on the rich experience accumulated in unmanned stores, we plan to accelerate our entry into overseas markets as well as the domestic market," said Yang Yoon-ji, CEO of Shinsegae I&C. "Going forward, we will develop it into an expandable solution that supports customer service, product guidance, and store operations by combining large language model (LLM)-based agent functions."