Related Experiment Video
Updated: Oct 2, 2025
14:27
Identification of Disease-related Spatial Covariance Patterns using Neuroimaging Data
Published on: June 26, 2013
15.8K
Revisiting Local Descriptors via Frequent Pattern Mining for Fine-Grained Image Retrieval
Min Zheng1, Yangliao Geng1, Qingyong Li1
1Beijing Key Lab of Traffic Data Analysis and Mining, Beijing Jiaotong University, Beijing 100044, China.
Entropy (Basel, Switzerland)
|February 25, 2022
Summary
This study introduces a new method for fine-grained image retrieval by combining global and local features. The approach enhances the ability to distinguish between similar classes, improving search accuracy.
Area of Science:
- Computer Science
- Artificial Intelligence
- Machine Learning
Background:
- Fine-grained image retrieval is challenging due to subtle inter-class differences and significant intra-class variations.
- Existing methods struggle when relying solely on global or local image features.
Purpose of the Study:
- To propose a novel method for fine-grained image retrieval that learns a global-local aware feature representation.
- To enhance the discriminative power among fine-grained classes for improved retrieval accuracy.
Main Methods:
- Extracting global features by selecting relevant deep descriptors.
- Utilizing frequent pattern mining to uncover relationships between image parts for representative local features.
- Designing an aggregation feature to integrate global and local information for a comprehensive representation.
Main Results:
- The proposed global-local aware feature representation significantly enhances the discriminative ability of fine-grained classes.
- Experimental results on five benchmark datasets confirm improved performance in fine-grained image retrieval.
Conclusions:
- The developed global-local aware feature representation is effective for fine-grained image retrieval.
- This approach addresses key challenges in fine-grained retrieval by leveraging both global context and local details.
