AI Sensor Fusion for Self-Checkout Missed Scan Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing self-checkout technologies suffer from high costs, susceptibility to errors, and inefficiencies in detecting missed scans, particularly due to issues with weight detection, RFID tags, eTag sensing, image recognition, and barcode identification, leading to economic losses and incorrect inventory records.
Innovation Solution
A self-checkout system integrated with sensors and artificial intelligence that uses multiple independent models for gesture and object recognition, combined with a checkout assistance decision model to determine missed scan events, and a kit for upgrading traditional checkout devices to include image sensors and control modules for self-checkout functionality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If weight-sensing method is used to identify goods, then the system can detect items on the checkout scale, but identification errors occur for different items with close weights and multiple items placed simultaneously
Solution Approach 1:
The patent combines weight-sensing technology with image recognition technology to create a hybrid detection system. The weight sensor provides initial item detection while the image recognition system captures and analyzes images of items on the scale, using AI algorithms to identify multiple items simultaneously and accurately, even when they have similar weights or are placed at the same time.
2Measurement precision
If RFID tag sensing technology is used to identify goods, then accurate identification is achieved, but the cost increases significantly
Solution Approach 1:
The patent replaces expensive RFID tags with barcode labels that can be printed on regular paper or adhesive materials. The barcode scanning system uses affordable image sensors and AI-based recognition algorithms to read barcodes, achieving accurate item identification without the high costs associated with RFID tag infrastructure and hardware.
3Adaptability or versatility
If eTag tag sensing technology is used to store asset information, then asset tracking is enabled, but investment in specific equipment and infrastructure is required
Solution Approach 1:
The patent uses standard barcode labels that can be applied to any item regardless of its value or type, eliminating the need for specialized eTag infrastructure. The same image sensor and AI recognition system that handles visual item identification also reads barcodes, providing multi-functional capability without requiring separate asset tracking equipment.
4Extent of automation
If pure image recognition technology is used to identify goods, then visual identification is achieved, but the technology is sensitive to ambient lighting and placement angle
Solution Approach 1:
The patent incorporates preprocessing steps that prepare images for recognition by adjusting lighting conditions, enhancing contrast, and normalizing image quality before AI analysis. The system captures multiple images if needed and applies image enhancement algorithms to ensure optimal recognition conditions, reducing sensitivity to ambient lighting variations and placement angles.
5Productivity
If barcode identification technology is used to check out goods, then item identification is accomplished, but scanning inaccuracies and missed scans occur frequently
Solution Approach 1:
The patent implements a feedback mechanism where the system continuously monitors the checkout process using image recognition. When an item is placed on the scale, the system captures an image, identifies the item, and verifies whether it has been scanned. If a missed scan is detected, the system provides real-time feedback to the customer through visual or audible alerts, prompting them to complete the scanning process.
6Ease of manufacture
If traditional checkout devices are upgraded to self-checkout systems, then construction costs are reduced, but existing devices lack self-checkout capabilities
Solution Approach 1:
The patent transforms traditional checkout devices into self-checkout systems by adding image sensors, barcode scanners, and AI-based recognition software. The upgraded system enables customers to independently perform item identification, scanning, and payment functions without requiring cashiers, while utilizing the existing hardware infrastructure of traditional checkout devices.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention relates to a self-checkout method (600). The method (600) includes: executing (601) following steps through a control module: activating (602) a first image sensor to capture a first image that includes a hand; activating (604) a second image sensor to capture a second image that includes an item; running (610) a gesture recognition independent model to generate a first recognition result based on the first image; running (611) an object recognition independent model to generate a third recognition result based on the second image; and executing and inputting the first recognition result and the third recognition result into a checkout assistance decision independent model to determine whether a missed scan event has occurred.