Self-Checkout Product Tracking via Image Rearrangement Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing self-checkout systems face challenges in accurately managing products when customers return them to different locations than where they were taken, leading to difficulties in maintaining a proper user article list and determining product movements during payment processes.
Innovation Solution
A self-checkout system that includes change detection, rearrangement detection, and shopping list generation mechanisms, using captured images to identify changes in product display states and associate them with person movements, allowing for accurate registration and deletion of products from shopping lists based on shelving information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If a system uses imaging devices to identify customer hands and track product movements, then automatic product identification is improved, but installation cost increases
Solution Approach 1:
The patent uses image capture devices to create visual copies of products and their locations on shelves. By processing these image copies rather than requiring complex physical tracking systems, the system achieves automatic product identification at lower cost. The image processing system creates digital representations of product movements and locations, replacing the need for expensive specialized imaging devices.
Solution Approach 2:
The patent replaces complex mechanical tracking systems with image processing and computer vision algorithms. Instead of using specialized imaging devices that require complex installation, the system uses standard cameras and image processing techniques to detect product movements, identify products, and track customer actions, significantly reducing installation costs while maintaining automation.
2Productivity
If a system tracks product flow lines to identify candidate products for payment, then payment process efficiency is improved, but the system cannot detect product returns to different locations
Solution Approach 1:
The patent implements a feedback mechanism that continuously monitors product locations and updates the system state based on detected movements. When a product is moved or returned to a different location, the image processing system detects the change and provides feedback to update the shopping list and payment process. This continuous feedback loop ensures both payment efficiency and accurate location tracking, resolving the contradiction by making the system responsive to actual product movements.
Solution Approach 2:
The patent makes the product tracking system dynamic by continuously updating product locations and associations based on real-time image analysis. Instead of static flow line assignments, the system dynamically adjusts product-customer associations when movements are detected. This dynamic approach allows the system to maintain payment efficiency while accurately detecting product returns to different locations, as the associations are updated in real-time based on actual product positions.
3Difficulty of detecting and measuring
If a system uses background subtraction methods to detect product changes, then product movement detection is improved, but the system cannot distinguish between product returns to different locations
Solution Approach 1:
The patent adds the dimension of location information to product movement detection. Instead of only detecting that a product moved, the system uses image processing to determine where the product moved to by analyzing the spatial coordinates in the captured images. This dimensional enhancement allows the system to distinguish between returns to the same location versus different locations, preserving location information while maintaining movement detection capability.
Solution Approach 2:
The patent performs preliminary registration of product locations and shelving information before detection begins. By establishing a baseline map of where products should be located and creating initial background images of product positions, the system can compare subsequent images against this preliminary data. This preliminary action enables the system to detect not only that products moved but also to identify specific location changes, preventing information loss while maintaining detection sensitivity.
Data Source
AI summary
On the basis of a detected change in the display state of the product and of a person included in the captured image or a person whose in-store flow line has been detected, a rearrangement detection means 820 detects that a product has been returned to a different location than the location from which the product was taken. A shopping list generation means 830 uses shelving information on a product shelf to specify a product for which there has been detected a change in the display state that is a result of the person having picked up the product, performs a registration process for registering the specified product on a shopping list that corresponds to the person, and performs a deletion process for deleting the product, which has been returned to the different location than the location from which the product was taken, from the shopping list.