Image Classification via Global and Local Feature Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image classification methods suffer from low accuracy due to the lack of variety in image features used, leading to inadequate classification results.
Innovation Solution
The method involves extracting global and local features from an input image, determining a local critical area, and obtaining correlation features to enhance the classification process, thereby enriching the feature information used for classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If only global features are extracted for image classification, then the classification process is simple and fast, but the classification accuracy is low due to lack of feature variety
Solution Approach 1:
The patent segments the image into multiple regions and extracts both global features from the entire image and local features from specific regions. This segmentation allows the system to capture diverse feature information at different scales, resolving the contradiction between simple classification and high accuracy by introducing multi-scale feature extraction without overwhelming complexity
Solution Approach 2:
The patent transitions from extracting only global features to extracting features at multiple dimensions (global and local levels). By adding the local feature dimension to the existing global feature dimension, the system enriches feature variety and improves classification accuracy while maintaining manageable complexity through structured feature integration
2Measurement precision
If multiple types of features (global and local) are extracted, then feature variety is enriched and classification accuracy improves, but the computational time and processing complexity increase
Solution Approach 1:
The patent applies local quality by extracting local features only from specific critical regions of the image rather than uniformly from the entire image. This selective approach enriches feature variety where needed while minimizing unnecessary computation in less informative regions, thus balancing accuracy improvement with processing time efficiency
Solution Approach 2:
The patent merges global features and local features into a unified feature representation for classification. By combining these feature types in an integrated manner, the system achieves high classification accuracy through feature variety while avoiding redundant computations and optimizing processing time through efficient feature fusion
Data Source
AI summary
The present disclosure describes a method, device, and non-transitory computer readable storage medium for classifying an input image. The method includes extracting, by a device, a global feature of a input image. The method includes determining, by the device, a local critical area of the input image based on the input image and the global feature; and extracting, by the device, a local feature of the local critical area. The method includes obtaining, by the device, a correlation feature of the input image based on the global feature of the input image and the local feature of the local critical area of the input image. The method includes classifying, by the device, the input image to obtain a classification result based on the global feature of the input image, the local feature of the local critical area, and the correlation feature of the input image.