Image Analysis Using Independent ROI Feature Extraction Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image analysis methods using machine learning face limitations in accuracy and objectivity, particularly when dealing with complex images or noise, as they either rely on entire images for classification, which can be unreliable, or focus on specific regions of interest, neglecting other important features.
Innovation Solution
An image analysis method that extracts regions of interest from a query image, applies independently learned ROI feature extraction models to calculate features, and uses a pre-learned integration analysis model to identify the image, thereby increasing accuracy and objectivity by separately detecting and reflecting features of main regions of interest.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the entire image is used as the learning target for classification, then all features in the image are considered, but the classification reliability deteriorates when the image contains complications or noise
Solution Approach 1:
The patent segments the image into multiple regions of interest (ROIs) based on object detection results. Each ROI is processed independently through separate feature extraction models, and the results are integrated for final classification. This segmentation approach isolates important features from noise and complications in other parts of the image, thereby improving classification reliability while managing image complexity.
2Measurement precision
If regions of interest are extracted and analyzed separately, then classification accuracy increases, but features from other regions are neglected
Solution Approach 1:
The patent merges the classification results from multiple independent ROI analysis models through an integration module. Each ROI is analyzed separately to capture local features with high precision, and then the results are combined to form the final classification decision. This merging process ensures that important features from different regions are all considered, preventing feature information loss while maintaining high classification accuracy.
3Measurement precision
If independently learned ROI feature extraction models are used for each region, then feature extraction accuracy improves, but the system complexity increases
Solution Approach 1:
The patent employs multiple independent feature extraction models that are each trained specifically for extracting features from different types of regions of interest. These specialized models serve universal purposes by handling various ROI types (such as different object categories or image regions) with high accuracy. The modular architecture allows each model to be independently optimized while contributing to the overall system functionality, balancing feature extraction accuracy with manageable system complexity.
Data Source
AI summary
The present disclosure relates to an image analysis method, system, and computer program. The image analysis method of the present disclosure includes: receiving a query image; extracting one or more regions of interest from the query image; calculating a first feature for each of the regions of interest by respectively applying the regions of interest to one or more ROI (region of interest) feature extraction models independently learned in order to extract features of the regions of interest; and calculating analysis values of the query image by applying the first features of the regions of interest to a pre-learned integration analysis model. According to the present disclosure, it is possible to reduce the influence on an analysis model by an error that training data created for map learning of an entire image may have, and it is also possible to increase learning accuracy and objectivity of a deep neural network.