top of page
Search

Product Attribution via Image Recognition (PAIR)

Scott Proost, Solutions Architect


Enter with us into the dynamic realm of the grocery industry, where products evolve constantly amidst a diverse array. Traditionally, manual business processes meticulously maintained catalogs and their attributes. However, these methods prove inadequate when faced with the frequency and scale of changes required to manage a comprehensive catalog. This is the backdrop against which Commerce Architects was approached by a leading grocery retailer facing just such a dilemma. 



Our Objective: 

Leverage Image Recognition and cutting-edge AI/ML technologies to enhance product attribution efficiency through automation.  

By doing so, we aimed to significantly elevate the quality and accuracy of data within the product catalog, thereby enriching the consumer shopping experience with up-to-date and precise product details. This enhancement enables improved search functionality, categorization, and faceting and offers numerous opportunities to personalize product preferences for individual shoppers.


To achieve this objective, we meticulously evaluated technology offerings and tools from leading AI vendors, such as GCP and AWS, to construct a flexible architecture. Our architecture incorporated a blend of:

  •  AWS infrastructure 

  • AWS Rekognition 

  • Components of GCP Vertex AI (VQA, Vision, LLM).  


With this foundation in place, we further refined the accuracy of our solution by optimizing interactions with the Large Language Model (LLM) through prompt engineering and tuning. To complement our solution, we developed a React UI to guide users through simplified business process workflows.




See PAIR in Action!



The outcomes of our Proof of Concept (POC) solution were overwhelmingly positive, underscoring the efficacy and efficiency of leveraging emerging AI technologies to address intricate business challenges. Converting product images into precise product data, coupled with optimizing manual business processes, represented a significant achievement. However, the POC merely scratched the surface, revealing numerous additional automation opportunities, such as product catalog auditing and data quality assessments.


As the technological landscape evolves at a rapid pace, Commerce Architects stands resolute in our dedication to investing significant time and effort into pioneering the application of AI/ML tools to solve real-world business challenges. Our team's deep technical expertise reflects our unwavering commitment to leveraging the most effective tools available, regardless of vendor, technology stack, or industry trends. We are steadfast in our mission to fulfill our brand promise: The right solution, on time, as promised.


Join us in shaping the future of business solutions—let's innovate together.


bottom of page