Abstract:In the field of image set classification, most existing works focus on exploiting effective latent discriminative features. However, it remains a research gap to efficiently handle this problem. In this paper, benefiting from the superiority of hashing in terms of its computational complexity and memory costs, we present a novel Discrete Metric Learning (DML) approach based on the Riemannian manifold for fast image set classification. The proposed DML jointly learns a metric in the induced space and a compact Hamming space, where efficient classification is carried out. Specifically, each image set is modeled as a point on Riemannian manifold after which the proposed DML minimizes the Hamming distance between similar Riemannian pairs and maximizes the Hamming distance between dissimilar ones by introducing a discriminative Mahalanobis-like matrix. To overcome the shortcoming of DML that relies on the vectorization of Riemannian representations, we further develop Bilinear Discrete Metric Learning (BDML) to directly manipulate the original Riemannian representations and explore the natural matrix structure for high-dimensional data. Different from conventional Riemannian metric learning methods, which require complicated Riemannian optimizations (e.g., Riemannian conjugate gradient), both DML and BDML can be efficiently optimized by computing the geodesic mean between the similarity matrix and inverse of the dissimilarity matrix. Extensive experiments conducted on different visual recognition tasks (face recognition, object recognition, and action recognition) demonstrate that the proposed methods achieve competitive performance in terms of accuracy and efficiency.

Semantic Discriminative Metric Learning for Image Similarity Measurement

Design and Application of an Image Classification Algorithm Based on Semantic Discrimination

Intermediate Semantics Based Distance Metric Learning For Video Annotation And Similarity Measurements

Semantic preserving distance metric learning and applications

Distance Metric Learning from Uncertain Side Information with Application to Automated Photo Tagging

Collaborative Similarity Metric Learning for Semantic Image Annotation and Retrieval.

Sparse semantic metric learning for image retrieval

Learning Based Neural Similarity Metrics for Multimedia Data Mining.

Similarity Measure Learning in Closed-Form Solution for Image Classification.

Semi-supervised LDA Based Method for Similarity Distance Metric Learning.

Distance Metric Learning For Semantic Segmentation Based Graph Hashing

High-Rankness Regularized Semi-Supervised Deep Metric Learning for Remote Sensing Imagery

Class-specific discriminative metric learning for scene recognition

Metric Learning with Feature Decomposition for Image Categorization

Enhanced Semantic Similarity Learning Framework for Image-Text Matching

Discrete Metric Learning for Fast Image Set Classification

Metric learning with geometric mean for similarities measurement

Semantic similarity metrics for image registration

Metric Learning By Similarity Network For Deep Semi-Supervised Learning

Learning Pseudo Metric for Intelligent Multimedia Data Classification and Retrieval

Self-weighted Discriminative Metric Learning Based on Deep Features for Scene Recognition