ProductEAI
An edge based computer vision system that reads store shelves in real time — spotting gaps, misplacements, and stock-outs before they cost a sale.
ProductEAI looks at a shelf the way an experienced store manager would — instantly.
Using a YOLO detector paired with a Vision Transformer classifier, it identifies products, flags empty facings, and catches misplaced items, turning a phone photo into actionable retail insight.
Strategized and trained on a vast inventory of data and models of different sizes to run on edge devices and on server.
Built as a research prototype and scaled for production, the model is currently delivering 96% accuracy for our retail giant.
Transitioning from barcode scans to computer vision–powered product intelligence, significantly reduced operational effort from ~4 hours to ~1.5 hours per cycle.
