Computer Vision Profile for Retail Transaction Discrepancy Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Retail stores face significant losses due to unpaid products, as customers often forget or intentionally fail to scan items at self-checkout, leading to unnoticed discrepancies during the checkout process.
Innovation Solution
A system utilizing multiple cameras to create a computer vision profile of items in a shopping cart, which is compared to an electronic transaction record, allowing for real-time detection and resolution of discrepancies through alerts or automated retrieval of unpaid items, using a network of local and central control circuits and databases.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If self-checkout is used to improve customer service and reduce labor costs, then ease of operation is improved, but loss of substance increases due to unscanned items
Solution Approach 1:
An intermediary system comprising cameras, processors, and databases is introduced between the customer and the checkout process. The system automatically captures images of items, identifies them through image processing, and compares them against transaction records, thereby preventing loss without requiring additional customer action or store employee intervention.
Solution Approach 2:
The manual mechanical process of scanning items by customers or employees is replaced with an automated optical system. Cameras capture images of items in the shopping cart, image processing algorithms automatically identify the items, and the system compares this data against transaction records, eliminating the need for physical scanning while preventing shrinkage.
2Measurement precision
If traditional scanning methods are used to detect unpaid items, then measurement precision is improved, but device complexity increases due to multiple cameras and processing systems
Solution Approach 1:
Instead of using complex scanning hardware, the system creates optical copies (images) of the items in the shopping cart. These images are then processed through automated recognition algorithms that identify items and compare them against transaction data, achieving accurate detection with a simpler overall system architecture.
Solution Approach 2:
The camera system serves multiple functions: it captures images for item identification, provides visual records for transaction verification, and enables automated comparison with purchase data. This multi-functionality reduces the need for separate specialized devices, thereby reducing overall system complexity while maintaining high detection precision.
3Loss of substance
If real-time comparison of items is performed to reduce unpaid losses, then loss of substance is reduced, but loss of time increases due to processing images and data
Solution Approach 1:
The system performs preliminary actions by continuously capturing images of items in the shopping cart and pre-processing this visual data. The images are stored and prepared for comparison before the customer completes checkout, so that when the transaction is finalized, the comparison can be quickly executed against the transaction record, minimizing actual processing time at the point of sale.
Solution Approach 2:
The image capture and preliminary processing occur continuously throughout the shopping experience rather than being triggered only at checkout. This continuous action ensures that when comparison is needed, the data is already prepared, significantly reducing the time required for actual verification while maintaining continuous monitoring to prevent unpaid losses.
Data Source
AI summary
A transaction record is created showing a purchase transaction of a customer. A CV profile showing a list of items in the transaction obtained from images is also obtained. The items in the transaction record are compared to items on the list. When there is a discrepancy, an action to take is determined.