Abstract:Significant progress has been made in instance image retrieval based on deep feature aggregation. However, existing approaches are limited by two issues: 1) The inability of deep features to localize target objects generates inaccurate feature descriptions and 2) using short vector feature representations provides unsatisfactory retrieval performance. To address these issues, we propose the compact deep semantic correlation descriptor (DSCD) approach, which has three main highlights: (1) Unlike manual labeling approaches, we propose a channel semantic correlation learning method to localize target objects. This method can learn semantic correlation information between channels from an auxiliary dataset and use this information as a priori knowledge to locate target objects well. (2) We propose a hierarchical attention mechanism to integrate multi-level features within target regions. It utilizes object and focus attention modules to capture global and local features, and constructs channel attention coefficients to enhance channels containing important target features. (3) We propose a general, yet efficient dimensionality reduction method named adaptive PCA-whitening to improve short vector feature retrieval performance. This method can dynamically fuse the feature vectors of different dimensions via adaptive coefficients, yielding a more compact and robust representation. Extensive experiments on six benchmark datasets show that our method achieves better retrieval performance compared to existing state-of-the-art unsupervised methods. Compared with the next-best unsupervised methods, the mAP scores of our method (dim = 128) are 11.2 %, 6.1 %, 9.3 %, 8.3 %, 14.3 %, and 22.1 % higher on the Oxford5K, Paris6K, Oxford105K, Paris106K, ROxford, and RParis datasets, respectively.

Image retrieval based on semicircle local binary patterns structure correlation descriptor

Enhanced Micro-structure Descriptor based Image Retrieval.

SIFT-Based Image Retrieval Combining the Distance Measure of Global Image and Sub-Image

SemSCo: Semantic Frequency Domain Scan Context for LiDAR-Based Place Recognition.

Image Retrieval Based on Structured Local Binary Kirsch Pattern.

Local Spatial Binary Pattern: a New Feature Descriptor for Content-Based Image Retrieval

Robust Local Feature Descriptor for Multisource Remote Sensing Image Registration

Texture Clustering Combining Complex Wavelet Features and Local Binary Pattern

Multi-Scale Local Spatial Binary Patterns for Content-Based Image Retrieval

Incorporating Spatial Distribution Feature with Local Patterns for Content-Based Image Retrieval

Local Ternary Cross Structure Pattern: A Color LBP Feature Extraction with Applications in CBIR

A new texture direction feature descriptor and its application in content-based image retrieval

An approach of content-based image retrieval based on image salient region

An Image Retrieval Combining Color and Spatial Information

Local Parallel Cross Pattern: A Color Texture Descriptor for Image Retrieval

Image retrieval using both color and local spatial feature histograms

Color-related Local Binary Pattern: A Learned Local Descriptor for Color Image Recognition

Efficient and distinctive binary descriptor for rotated circular image recognition

Image retrieval using compact deep semantic correlation descriptors

A Novel Feature Descriptor Based Local Information for Image Retrieval

Image Retrieval Approach Based on Sparse Canonical Correlation Analysis