Object Recognition Apparatus Candidate Filtering for Retail
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In object recognition systems, especially in store accounting systems, the efficiency of identifying products decreases when there are multiple similar products registered for sale, as the system needs to compare feature values of captured images with a large number of candidates, leading to increased processing time and user confusion.
Innovation Solution
An apparatus and method for managing data used in object recognition that captures images, stores product data, and allows user selection to invalidate similar products, reducing the number of candidates and improving recognition efficiency by displaying only valid products on an operation panel.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple similar products are registered for sale in the object recognition system, then the system can recognize a wider variety of products, but the efficiency of identifying the correct product decreases due to increased comparison time and user confusion
Solution Approach 1:
The patent segments the product database by storing validity information for each registered product. The system divides the recognition process into two stages: first identifying all similar products based on image features, then filtering to show only valid (currently sold) products to the user. This segmentation resolves the contradiction by maintaining comprehensive product coverage while improving identification efficiency through selective display.
Solution Approach 2:
The system performs preliminary filtering of products by validity status before presenting candidates to the user. By pre-processing the product list to exclude invalid products, the system reduces the number of comparisons users must evaluate, thereby improving identification efficiency without reducing the overall product recognition capability.
2Adaptability or versatility
If the system compares captured image feature values with a large number of registered product candidates, then more products can be identified, but the processing time increases
Solution Approach 1:
The patent extracts and separates the validity information from the product database, storing it as distinct data. During recognition, the system extracts only the valid products from the full database for comparison and display. This extraction principle reduces processing time by eliminating comparisons with invalid products while maintaining the capability to identify all product types.
Solution Approach 2:
The system performs preliminary filtering to separate valid and invalid products before the recognition comparison process. By pre-organizing the database with validity flags and using them to filter candidates, the system reduces the number of comparisons required while maintaining comprehensive product identification capability.
3Adaptability or versatility
If the system displays all similar products as candidates, then users have more options to choose from, but the ease of operation decreases due to user confusion among multiple candidates
Solution Approach 1:
The system extracts and displays only the valid products from the set of similar products identified through image recognition. By separating valid candidates from invalid ones and presenting only the relevant subset to users, the system maintains flexible product selection capability while significantly improving ease of operation by reducing candidate confusion.
Solution Approach 2:
The system performs preliminary filtering of candidate products by validity status before presenting them to users. This pre-processing step removes invalid products from the candidate list, allowing users to make selections more easily without being confused by products that are no longer available, while still providing access to all currently sellable options.
Data Source
AI summary
An apparatus for managing data used for object recognition includes an image capturing unit configured to capture an image of an object, a storage unit storing image data of products registered for sale, an operation panel configured to receive a user selection, and a processor configured to determine the products registered for sale that are similar to the object, based on the captured image and the image data, display the similar products on the operation panel as user selectable items, determine one or more of the similar products designated by the user selection, and invalidate the designated similar products so that efficiency of object recognition by the apparatus can be improved.