Shelf Planogram Stitching for Accurate Retail Inventory Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Inventory management in retail environments is time-consuming and error-prone, with current methods like handheld barcode scanning being slow and inefficient, and existing computer vision approaches having low accuracy in tracking items across images and accounting for relative positioning.
Innovation Solution
A system and method using computer vision techniques to create planograms from stitched shelf images, employing machine learning models to identify items and their relative positions, allowing for rapid and accurate inventory updates and navigation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If handheld barcode scanning is used for inventory management, then items can be tracked, but the process is time-consuming and slow
Solution Approach 1:
The patent replaces manual barcode scanning with computer vision technology using cameras and machine learning models to automatically detect and track items on shelves. This substitution of mechanical scanning with optical recognition systems enables simultaneous capture of multiple items, dramatically improving inventory update speed while maintaining accuracy through advanced image processing algorithms.
2Productivity
If computer vision approaches are used to track items, then speed can be improved, but accuracy in tracking items across images and accounting for relative positioning deteriorates
Solution Approach 1:
The patent introduces planograms as intermediary reference data that contain predefined item positions and relationships. By comparing detected item positions in images against planogram references, the system accurately determines relative positioning and identifies items even when perspectives or lighting conditions vary, thereby maintaining high positioning accuracy while enabling rapid processing of multiple images.
Solution Approach 2:
The system uses planogram data as feedback to continuously refine item identification and positioning accuracy. By iteratively comparing detected items against expected planogram configurations, the system can correct positioning errors and improve tracking accuracy across multiple images while maintaining high processing speed.
3Quantity of substance
If multiple images of shelf portions are captured, then complete inventory coverage is achieved, but processing complexity increases
Solution Approach 1:
The patent divides the shelf into multiple imageable portions and processes each portion separately using individual images or image segments. This segmentation allows the system to handle large shelf areas by breaking them into manageable sections, reducing the computational complexity of processing entire shelves at once while ensuring complete inventory coverage through systematic coverage of all segments.
Data Source
AI summary
A method includes receiving, by a computer, a plurality of images of portions of a shelf unit from one or more user devices. Each image captures a different portion of the shelf unit. The method also includes creating, by the computer, a planogram of items on the shelf unit using image data from the plurality of images of the portions. The method also includes additional processing using the planogram.