Simulated Card Image Training for Subtype Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems fail to accurately identify card types and their subtypes in electronic wallets, especially when new subtypes are introduced, as they rely on manual entry and are not robust enough to handle variations in card designs and information formats.
Innovation Solution
A computer model, such as a convolutional neural network, is trained using simulated card images with identifying markings to anticipate various subtypes, allowing for the identification of card types even when new subtypes are not available during training, by integrating these markings into the training set and applying preprocessing techniques to standardize card images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a computer model is trained only with existing card subtype examples, then it can accurately identify known subtypes, but it fails to correctly identify new subtypes when they are issued by an entity
Solution Approach 1:
The system performs preliminary actions by generating simulated card images that represent potential future subtypes before they actually appear in the training data. By anticipating possible card designs and variations in advance, the model is pre-trained to recognize patterns it hasn't explicitly seen, enabling it to adapt to new subtypes when they are introduced without requiring retraining
Solution Approach 2:
The system creates copies of existing card images through simulation, generating synthetic representations of potential card subtypes. These simulated copies incorporate the identifying markings and design patterns of known card types while varying other features to represent potential future variations, allowing the model to learn from these artificial copies rather than requiring actual physical cards of every possible subtype
2Ease of operation
If manual entry methods are used for card information, then users can enter any card information, but the process is time-consuming and not efficient for populating electronic wallets
Solution Approach 1:
The system enables self-service by allowing the computer model to automatically capture, analyze, and extract card information from images without requiring user intervention for data entry. The model independently identifies card types, extracts relevant information, and populates the electronic wallet, making the system serve itself rather than requiring manual user input while maintaining flexibility in handling various card formats
3Loss of information
If character recognition is used to analyze card images, then numbers and basic information can be extracted, but the system cannot reliably identify card types and associated entities
Solution Approach 1:
The computer model performs multiple functions simultaneously: it extracts numerical data and text information through recognition, identifies card types by analyzing visual patterns and layout, and determines associated entities by recognizing identifying markings. This multi-functional approach allows the system to handle various card formats and information types with a single unified model, achieving both data extraction and accurate card type identification
Data Source
AI summary
A computer model to identify a type of physical card is trained using simulated card images. The physical card may exist with various subtypes, some of which may not exist or be unavailable when the model is trained. To more robustly identify these subtypes, the training data set for the computer model includes simulated card images that are generated for the card type. The simulated card images are generated based on a semi-randomized background that varies in appearance, onto which an identifying marking of the card type is superimposed, such that the training data for the computer model includes additional randomized sample card images and ensure the model is robust to further variations in subtypes.