Product authentication as a service runs AI-algorithm on close-up images
For a monthly subscription fee and a one-time setup and training fee (including a dedicated handheld microscope that takes close-up pictures with a magnification of 260x), the company runs any newly acquired images by a deep learning-driven database with millions of microscopic images to authenticate luxury goods.
Using a dedicated app on their smartphone, customers get the authentication results in real-time and with more accurate results than what resellers can provide with a manual and subjective evaluation process.
Launched in 2016, Entrupy’s patented technology is currently used by hundreds of secondary retailers and marketplaces to authenticate handbags and wallets from brands including Louis Vuitton, Chanel and Hermès. To date, Entrupy has authenticated $14 million worth of goods and it guarantees the veracity of its findings with the Entrupy Certificate of Authenticity, backed by a financial guarantee.
“For the moment, we are servicing the authentication needs for businesses. Our customers range from pawn establishments to large online resellers and marketplaces. In the future, we intend to extend the service to other segments of the market” commented Vidyuth Srinivasan, CEO and co-founder of Entrupy when contacted by eeNews Europe.
Revealing more about the mode of authentication, Srinivasan wrote: “The microscope provides a magnification of over 260x the human eye and we require around 24 microscopic samples to provide accurate results. As we grow, we plan to reduce the number of images while still maintaining detection accuracy. Depending on the item, it could take a few days to a couple months to register a new item. We spent a while building our database to play catch-up with a lot of the brands. We have data samples across 80 years for some of the brands, so that took time to build”.
Then any subsequent verification shots taken by customers during a product examination are added to the database when the item is identified as genuine, increasing the diversity of data and making future results more robust.
“Most of the algorithms we use are custom-coded to cater to specific optical features. In some cases, we use many more vision and machine learning techniques outside of Deep Learning, so it is an amalgamation of different systems that provide consistent and accurate results” Srinivasan concluded.
Entrupy – www.entrupy.com
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
