AI-Powered Smart Inventory Management System for Retail Shops

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

Maintaining reliable stock information is a recurring challenge for small retail businesses, where inventory records are often updated through manual counting and periodic shelf inspections.These practices require regular effort and can produce inaccurate records when products are overlooked or their movement is not recorded.This paper presents a vision-based inventory monitoring system that uses object detection to estimate the number of products visible on a retail shelf.A webcam provides the visual input, and a YOLO-based detection model identifies objects in captured frames.The detected labels are processed by an inventory-management mechanism that compares observations across successive frames.A two-second stabilization interval is used before confirming product removal, reducing incorrect updates caused by temporary detection failures.Product additions, removals, and current inventory information are stored in CSV files.A dashboard developed with Dash presents inventory counts, the number of detected item categories, and stock changes.The implementation demonstrates how computer vision can support shelf-level inventory monitoring using a camera and a laptop-based processing environment.Custom product training, automated low-stock notifications, database integration, and demand prediction are possible extensions.