Self-Checkout Item Detection Using Computer Vision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Errors in the self-checkout process for weighing items, such as produce, can lead to lost sales or 'shrinkage' due to incorrect item identification, incomplete scanning, or improper weighing.
Innovation Solution
The implementation of an image recognition system and computer vision techniques within the point-of-sale (POS) system to detect errors in weighing and scanning by analyzing images of the items being weighed and comparing them to the data input by the user.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If manual item identification and weighing is used at self-checkout stations, then customers can independently complete purchases, but errors in item identification and weighing occur leading to shrinkage
Solution Approach 1:
The system captures images of items on the weighing platter, processes them through computer vision to identify items and detect their placement, then provides feedback by comparing actual placement with expected placement. This feedback loop enables automatic correction of weighing errors while maintaining self-checkout operation.
Solution Approach 2:
The patent replaces manual mechanical weighing and visual inspection with an automated computer vision system that uses image processing and machine learning algorithms to identify items, detect their positions, and verify proper placement on the weighing platter, eliminating human error in the process.
2Device complexity
If customers manually place items on the scale, then the weighing process is simple, but items may be improperly positioned leading to inaccurate weight measurements
Solution Approach 1:
The system performs preliminary detection by capturing an image of the item placement before finalizing the weight measurement. The computer vision system analyzes the image to determine if the item is properly positioned on the weighing platter, and only after verification does it proceed with accurate weight measurement and transaction processing.
Solution Approach 2:
The patent introduces an image capture device and computer vision processing as an intermediary between the physical item placement and the weight measurement. This intermediary layer verifies that items are properly positioned before the scale records the weight, ensuring measurement precision without adding complexity to the weighing mechanism itself.
3Device complexity
If traditional weighing systems are used, then the system is simple, but errors such as items not fully on the scale go undetected
Solution Approach 1:
The patent adds a visual dimension to the traditional weighing system by incorporating image capture devices that photograph items from above. This dimensional addition allows the system to detect whether items are fully on the weighing platter by analyzing their position in the image, providing information that the scale alone cannot detect.
Solution Approach 2:
The system creates a visual copy (image) of the item placement on the weighing platter and processes this copy through computer vision algorithms to detect improper placement. This copying approach allows verification of item position without physically interfering with the weighing process or adding mechanical complexity.
Data Source
AI summary
A method is provided. The method includes receiving image information corresponding to a point-of-sale (POS) transaction, the image information includes an item being weighed, receiving an indication of an item weight or an indication of an item identifier for the item associated with the POS transaction, determining an expected item weight or an expected item identifier for the item based on the image information, determining a condition exists between an item weight and the expected item weight or between an item identifier and the expected item identifier and managing the POS transaction based on the condition. A system is also provided.