Pensa Granted Patent for AI That Can Identify On-Shelf Retail Products at Scale While Maintaining High Data Accuracy
Innovation advances market leadership in analyzing large volumes of retail shelf data for CPG brands and retailers without degrading accuracy
Pensa Systems, a leading innovator in automated retail shelf intelligence, today announced it has been granted a U.S. patent that provides an unprecedented ability to automatically distinguish between large numbers of different products while maintaining the highest levels of data accuracy. Pensa previously announced it has visually recognized more than 3 billion CPG product images with higher than 98% accuracy.
Read More: SalesTechStar Interview with Jen Jackson, Vice President of Customer Experience at Dialpad
Pensa fills a gap for CPG brands and retailers by providing a continuous view of actual products on the retail shelf as an alternative to legacy approaches that estimate shelf inventory based on a combination of back-room inventory and point of sale data. These legacy approaches are highly inaccurate. Actual product stock-out rates can be up to 30% and are often 17 points worse than estimated stock-out rates. This discrepancy leads to empty shelves, unsatisfied customers, incorrect supply chain decisions, and ultimately reduced sales and profitability. The problem worsens dramatically in omnichannel operations resulting in high product substitution rates that negatively impact profitability and customer loyalty.
Pensa’s Artificial Intelligence (AI) is the first fully automated shelf intelligence solution. To automatically deliver accurate, near real-time shelf data, Pensa’s AI learns to visually recognize and distinguish between products on the shelf much as a human does versus scanning a barcode on the back of a package or comparing a single product image against a product image database. Pensa captures then analyzes a video stream of hundreds of images, taken from numerous angles, of each individual product on the shelf to accurately identify and distinguish between products in near real-time to calculate a highly accurate view of shelf inventory. See Pensa capture and analyze a store aisle in seconds here.
Read More: SalesTechStar Interview with Yaron Schechtman, SVP of Global Sales, Namogoo
The average grocery store carries over 30,000 products. Many are very similar in appearance, but in fact are distinct items in inventory. Historically, it has been very difficult for AI to identify products at any realistic scale without accuracy degrading. Pensa’s AI maintains accuracy to accommodate even the largest retail environments, which brings shelf-data accuracy at scale to brands and retailers for the first time. Pensa’s AI can distinguish between individual products without requiring training per store and without depending on inaccurate shelf maps for what products “should be” positioned on the shelf. Instead Pensa’s AI learns the products as it goes, imputing by itself how shelves are managed (what the industry refers to as a planogram) to automatically and accurately identify products that are running low or out of stock.
“We are pleased to have received this validation of the innovation we are developing to help us drive growth for our customers,” said Jim Dutton, CTO of Pensa Systems. “Helping customers grow while removing risk and complexity from their business is an important focus for us and delivering the highest quality shelf data at scale supports that focus. As consumers continue to demand more from brands and retailers, we will continue to develop and patent innovation to help the CPG industry meet these demands.”
