Multi-Category Image Classification via Binary Feature Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object classification models are inadequate for classifying objects into multiple categories, which is necessary for digital files containing multiple image categories, as they typically focus on single-category classification.
Innovation Solution
An image object classification method that involves extracting binary-classified characteristics, applying dimensionality reduction through concatenation, performing characteristics abstraction, and integrating results using a dot product of matrices to obtain a classification result, allowing for classification into multiple categories.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional single-category classification models are used, then the classification process is simple, but the model cannot classify objects into multiple categories which is needed for digital files containing multiple image categories
Solution Approach 1:
The patent divides the classification task into multiple independent binary classifiers, each handling a specific category. This segmentation allows the system to achieve multi-category classification capability while maintaining relatively simple individual classifier structures, resolving the contradiction between versatility and complexity.
Solution Approach 2:
The patent creates a universal classification framework that can handle multiple categories through a standardized binary classifier structure. The same basic classifier design is reused across different categories, achieving multi-functionality without proportionally increasing complexity.
2Measurement precision
If multiple binary-classified characteristics are extracted independently, then the classification accuracy for each category improves, but the result integration becomes complex
Solution Approach 1:
The patent merges multiple independent binary classification results into a unified multi-category classification output. By combining the results through systematic integration, the system achieves high classification accuracy across multiple categories while managing the complexity of result integration through a structured approach.
Data Source
AI summary
An image object classification method and system are disclosed. The method is executed by a processor coupled to a memory. The method includes: providing an image file including at least one image object, performing a process of extracting multiple binary-classified characteristics on the image object to obtain a plurality of first results independent of each other in categories, combining the plurality of first results in a manner of dimensionality reduction based on concatenation, performing a process of characteristics abstraction on the combined first results to obtain a second result, and performing a process of characteristics integration on the plurality of first results and the second result in a manner of dot product of matrices to obtain a classification result.