Matej started his journey at Microblink not so long ago, in January 2022. As a Machine learning engineer, he specializes in computer vision, particularly object detection and semantic segmentation.
When In-store Product Recognition kicked off as Microblink’s newest project, Matej and his colleagues took the challenge of initial ML research and developing the first deep learning models in the pipeline. Further research and development led to a practical end-to-end product recognition solution, thanks to the efforts of Microblink’s engineers from a broad area of technical expertise.
In-store product recognition with AI
Although recent years and the COVID-19 pandemic showed a significant increase in online shopping, most customers still do their daily grocery and necessities shopping in physical retail stores. Except for coupons or post-shopping future discounts, there is little opportunity for product manufacturing companies to communicate and offer discounts, bonuses, and other shopping incentives. On the other hand, manufacturers have no direct and real-time information about their product’s daily and weekly sales statistics to boost their marketing efforts and offer shopping incentives.
Microblink’s In-store Product Recognition pipeline solves those problems in a reliable yet innovative way. Product Recognition is running even the heaviest of computer vision tasks on mobile devices by combining machine learning models with some custom heuristics and highly optimized implementations. It offers the ease of use of a daily-performed task for us nowadays – pointing your phone into the shelf or individual product and voila, see the machine learning magic happen!