Mobile Computer Vision Shelf Mapping for Out-of-Stock Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing inventory management systems in retail environments face challenges in accurately and efficiently determining stock levels on shelves, often relying on manual counts, specialized equipment prone to malfunction, or server systems that fail to account for human variation, leading to inefficiencies and inaccuracies.
Innovation Solution
A mobile apparatus using computer vision and machine-learning classifiers integrated with movable structures like shopping carts, order picking carts, or cleaning devices to optically detect inventory conditions, combining low-resolution and high-resolution imaging for accurate, real-time stock level determination, and decentralized processing with edge computing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual inventory counts are performed, then inventory accuracy can be achieved, but labor expense and time consumption increase significantly
Solution Approach 1:
The patent replaces manual mechanical counting with an automated optical detection system using cameras and machine learning classifiers. The system captures images of shelves and automatically identifies inventory conditions, eliminating the need for human labor while maintaining high accuracy in determining stock levels.
Solution Approach 2:
The system enables self-monitoring of inventory levels through automated image capture and analysis. The mobile apparatus independently performs detection, classification, and reporting of inventory conditions without requiring human intervention, allowing continuous autonomous monitoring of stock levels.
2Reliability
If specialized shelf equipment with sensors is deployed, then real-time inventory detection capability is improved, but device complexity and implementation cost increase
Solution Approach 1:
The patent applies computer vision technology to multiple inventory detection scenarios including shelf monitoring, bin tracking, and stock level assessment. The same camera-based system can detect various inventory conditions across different retail environments, eliminating the need for specialized sensors for each specific application.
Solution Approach 2:
The system extracts only the essential function of inventory detection from complex sensor systems, using simple camera-based optical detection. By removing unnecessary specialized components and retaining only the core detection capability through image analysis, the system achieves real-time monitoring with reduced complexity.
3Measurement precision
If specialized shelf equipment is installed, then inventory detection accuracy is improved, but maintenance effort and malfunction risk increase
Solution Approach 1:
The patent uses inexpensive, readily available camera components instead of expensive specialized sensors. These standard camera devices are more easily replaced and maintained if needed, reducing the overall maintenance burden and cost while maintaining adequate detection accuracy for inventory monitoring purposes.
4Loss of information
If inventory tracking server systems are used, then data correlation capability is improved, but ability to account for human variation deteriorates
Solution Approach 1:
The system incorporates feedback loops where detected inventory conditions are continuously monitored and used to adjust detection parameters and improve accuracy. The machine learning classifiers are trained on diverse data including cases involving human variation, allowing the system to adapt and accurately detect situations such as items temporarily removed by customers or misplaced items.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables frequent, accurate, and unobtrusive inventory monitoring with reduced complexity and cost, utilizing existing devices for efficient data collection and processing, minimizing power consumption, and reducing network load.
Implementation Method 1
a camera in data communication with one or more processors... receiving image data from the camera
Data Source
AI summary
Disclosed are systems and techniques for determining out of stock conditions on shelves. The techniques can include receiving, by a computing system, image data from a camera having pixel locations that each uniquely address and store a pixel value, generating a backing map having cell locations that each uniquely address and share a unique address with a corresponding pixel location in the image data, each cell location storing a backing value being an empty value if the pixel value is classified as showing the backing of a shelf and the backing value being a nonempty value if the pixel value is classified as not showing the backing of the shelf, determining, in the backing map, a shelf area representing a location of the captured shelf, and identifying an empty area by finding an area above the shelf area containing a threshold number of cell locations storing the empty value.