Commodity Identification Using Mark Detection and Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing commodity identification systems incorrectly classify unregistered commodities as registered ones due to the lack of distinct markers, leading to inaccurate identification.
Innovation Solution
A commodity identification system using a mark detector and neural networks to identify commodities, where a first neural network determines the commodity group based on detected marks and a second neural network identifies individual commodities within those groups, preventing misclassification by excluding unregistered commodities from processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If image recognition is performed on taken images without mark detection, then commodity identification can be performed on all imaged objects, but unregistered commodities are incorrectly classified as registered commodities
Solution Approach 1:
The identification process is segmented into three distinct stages: mark detection, commodity group identification, and commodity identification. This segmentation allows the system to first verify if a commodity has a detectable mark before attempting identification, preventing misclassification of unregistered commodities while maintaining the ability to identify registered ones accurately
Solution Approach 2:
A mark detector serves as an intermediary component between the image input and the neural network-based identification system. This intermediary filters out unregistered commodities that lack detectable marks before they enter the identification pipeline, preventing erroneous classifications while allowing registered commodities to proceed through the identification process
2Device complexity
If a single neural network is used for commodity identification, then the system structure is simple, but the system cannot reliably distinguish between registered and unregistered commodities
Solution Approach 1:
The neural network functionality is segmented into multiple specialized networks: a mark detector for detecting commodity marks, a first neural network for identifying commodity groups, and a second neural network for identifying specific commodities. This segmentation of functional responsibilities improves reliability by ensuring that unregistered commodities without marks are correctly identified as such, while maintaining manageable system complexity through modular architecture
Solution Approach 2:
The mark detection process performs a preliminary action before the identification process begins. By first detecting whether a commodity has a mark and identifying its group, the system prepares the data in advance for the final identification stage, ensuring that unregistered commodities are excluded from the identification results while maintaining clear system structure
Data Source
AI summary
A commodity identification device is provided with: one or more processing devices; and one or more storage devices storing instructions for causing the one or more processing devices to: obtain a taken image; determine whether a commodity is provided with a corresponding mark or not by using a mark detector on the obtained taken image; identify which of a plurality of commodity groups a commodity provided with the mark belongs to by using a first neural network having learned the commodity, on the taken image where the mark is detected; and identify the commodity by using a second neural network having performed learning for each of the commodity groups.