Product Image Registration for Accurate Shelf Scanning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object scanning systems require tedious manual registration of learning images for each target product and are prone to inaccuracies when scanning products not currently on sale, necessitating cumbersome for-sale settings.
Innovation Solution
An information processing device with capture, extraction, designation reception, and registration means that automates the registration of learning images and for-sale settings by referencing product information storage, allowing easy product identification and sale management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual registration of learning images is performed for each target product, then image recognition accuracy is improved, but operation time and labor intensity increase significantly
Solution Approach 1:
The system performs self-learning by automatically capturing images of products on the shelf and registering them as learning images without requiring manual intervention. The image recognition device autonomously identifies products and accumulates learning data, eliminating the need for operators to manually register each learning image while maintaining recognition accuracy.
Solution Approach 2:
The system performs preliminary image capture and registration during idle periods or automatically in the background before actual scanning operations. By pre-capturing product images and registering them as learning images in advance, the system prepares recognition data without interrupting normal operations or requiring dedicated registration time.
2Adaptability or versatility
If all registered products are scanned, then product identification capability is improved, but irrelevant products not on sale are also extracted increasing operation complexity
Solution Approach 1:
The system extracts and displays only the relevant subset of products that are currently on sale from the complete registered product database. By separating saleable products from non-saleable ones and presenting only the necessary options to the operator, the system maintains comprehensive product identification capability while simplifying the scanning operation to show only relevant results.
3Measurement precision
If for-sale setting is performed manually for each product, then scanning accuracy is improved, but setup time and operational burden increase
Solution Approach 1:
The system automatically manages for-sale settings by detecting which products are currently available on the shelf and updating the saleable product list without requiring manual operator input. The image recognition device autonomously determines product availability and configures scanning parameters, maintaining scanning accuracy while eliminating manual setup operations.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
This information processing device comprises: an imaging means that captures an image of a product; an extracting means that uses the image of the product captured by the imaging means as a basis for extracting candidate products to which a training image is to be added by referring to a product information storage means storing product information; a designation reception means that receives a designation of a product to which the training image is to be added and registered from among the candidate products extracted by the extracting means; and a registering means that registers the image of the product captured by the imaging means as the training image in the product information storage means, in association with the product designated in the designation reception means.