Object Recognition System with Pre-Registration Similarity Check

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In commodity reading apparatuses, incorrect object recognition can occur when registering new commodities with feature values similar to existing ones, leading to potential misclassification during sales registration processing.

Innovation Solution

A commodity reading apparatus comprising a designation module, extraction module, storage module, calculation module, learning module, and notification module, which compares feature values and executes notification on recognition performance before learning processing, ensuring accurate registration by designating learning targets and calculating similarity degrees to prevent misclassification.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If the dictionary for collation is updated by registering a new commodity, then the recognition capability is improved, but the recognition accuracy deteriorates when the new commodity has similar feature values to existing commodities

Engineering Contradiction:
Improverecognition capabilityVSAvoidrecognition accuracy
Core Design Contradiction:
Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The system performs preliminary similarity assessment between the new commodity's feature values and existing commodity feature values before completing the registration. By calculating similarity degrees in advance and comparing them against threshold values, the system identifies potential recognition conflicts before they occur, allowing preventive actions to be taken.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system provides feedback to the user regarding the similarity degree between the new commodity and existing commodities in the dictionary. This feedback mechanism allows users to review the potential impact of registration on recognition accuracy and make informed decisions about whether to proceed with the registration.

Inventive Principle:
Principle #23Feedback

2Measurement precision

If feature values are compared and similarity degree is calculated before registration, then recognition accuracy is maintained, but the processing time increases

Engineering Contradiction:
Improverecognition accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system changes the parameter of similarity threshold values to balance between recognition accuracy and processing efficiency. By adjusting threshold values, the system can control the stringency of similarity checks, allowing faster processing when lower precision is acceptable and more thorough checks when higher precision is required.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If the system notifies users about similarity issues before registration, then recognition accuracy is preserved, but the registration process becomes more complex

Engineering Contradiction:
Improverecognition accuracyVSAvoidregistration process complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system provides self-service functionality by automatically calculating similarity degrees and generating notifications about potential recognition issues. Users can review the automated assessment results and make registration decisions based on clear, system-generated information, reducing the need for complex manual evaluation procedures.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP3293674B1Object recognation system and method of registering a new object
Publication Date: 2023.10.04 TOSHIBA TEC KK
  • EP3293674B1 patent drawingFigure 1
  • EP3293674B1 patent drawingFigure 2
  • EP3293674B1 patent drawingFigure 3

AI summary

A calculation module compares the first feature value extracted and a second feature value indicating feature of each commodity registered in a dictionary serving as a storage module to calculate a similarity degree therebetween. A notification module executes notification as to a recognition performance prior to a leaning processing in which the first feature value is registered as the feature value of the learning target article in the storage module by a learning module if a similarity degree between the first feature value and the second feature value of the registered commodities excluding the learning target article is equal to or greater than a first threshold value or a similarity degree between the first feature value and the second feature value of the learning target article contained in the registered commodities is equal to or smaller than a second threshold value smaller than the first threshold value.