Self-Service Checkout Fraud Detection via Zero-Shot Image Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Self-service checkout systems face challenges in detecting fraudulent activities, such as label switching and the 'banana trick,' due to the need for extensive training data and the frequent replacement of commodity products, which complicates the tuning of image recognition AI.
Innovation Solution
An alert generation program that utilizes a zero-shot image classifier and a hierarchical structure database to analyze video images from self-service checkout systems, identifying abnormal commodity product registrations and generating alerts for fraudulent activities without requiring extensive retraining.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If image recognition AI is used to detect fraudulent activities in self-service checkout systems, then detection capability is improved, but the need for extensive training data and frequent retraining increases
Solution Approach 1:
The patent introduces an intermediary system that captures images of commodity products at the checkout register and compares them against a database of registered product images. This intermediary image comparison mechanism bridges the gap between simple barcode scanning and complex AI analysis, enabling fraud detection without requiring extensive training data for AI models.
Solution Approach 2:
The patent creates a copy database of commodity product images that are registered in advance. Instead of training AI to recognize products from scratch, the system copies pre-registered product images and compares incoming images against these copies, eliminating the need for continuous AI retraining when products are replaced.
2Measurement precision
If traditional image recognition AI is deployed for fraud detection, then detection accuracy is improved, but adaptability to frequent product replacement deteriorates
Solution Approach 1:
The patent implements a dynamic image database that can be easily updated when commodity products are replaced. The system allows store employees to register new product images and automatically updates the comparison database, enabling the system to adapt to product lifecycle changes without retraining AI models. This dynamic update mechanism maintains high identification accuracy while being highly adaptable to frequent product replacements.
3Reliability
If comprehensive video analysis is performed on all checkout activities, then fraud detection capability is improved, but processing time and computational resources increase
Solution Approach 1:
The patent extracts only the essential image data from video footage - specifically, images of commodity products on the checkout register and in shopping bags. By extracting only these critical frames for comparison rather than analyzing entire video sequences, the system maintains high fraud detection capability while significantly reducing processing time and computational resource requirements.
Data Source
AI summary
A non-transitory computer-readable recording medium stores therein an alert generation program that causes a computer to execute a process including acquiring video image of a person who is scanning a code of a commodity product to an accounting machine, specifying, by analyzing the acquired video image, from among a plurality of commodity product candidates that are set in advance, a commodity product candidate that corresponds to the commodity product that is included in the video image, acquiring an item of the commodity product that has been registered to the accounting machine by scanning the code of the commodity product to the accounting machine, and generating, based on an item of the specified commodity product candidate and the item of the commodity product acquired from the accounting machine, an alert that indicates an abnormality of the commodity product that has been registered to the accounting machine.