Abstract:Feature learning is an intensively studied research topic in image classification. Although existing methods like sparse coding, locality-constrained linear coding, fisher vector encoding, etc., have shown their effectiveness in image representation, most of them overlook a phenomenon called thesmall sample size problem, where the number of training samples is relatively smaller than the dimensionality of the features, which may limit the predictive power of the classifier. Subspace learning is a strategy to mitigate this problem by reducing the dimensionality of the features. However, most conventional subspace learning methods attempt to learn a global subspace to discriminate all the classes, which proves to be difficult and ineffective in multi-class classification task. To this end, we propose to learn a local subspace for each sample instead of learning a global subspace for all samples. Our key observation is that, in multi-class image classification, the label of each testing sample is only confused by a few classes which have very similar visual appearance to it. Thus, in this work, we propose a coarse-to-fine strategy, which first picks out such classes, and then conducts a local subspace learning to discriminate them. As the subspace learning method is regularized and conducted within some selected classes, we term it selective regularized subspace learning (SRSL), and we term our classification pipeline selective regularized subspace learning based multi-class image classification (SRSL_MIC). Experimental results on four representative datasets (Caltech-101, Indoor-67, ORL Faces and AR Faces) demonstrate the effectiveness of the proposed method.

Local subspace analysis for astronomical object detection and star/galaxy classification

Improved System for Object Detection and Star/Galaxy Classification Via Local Subspace Analysis

Learning a Locality Preserving Subspace for Visual Recognition.

Face Recognition Using Data Driven Local Appearance Features

Improved Robust Kernel Subspace for Object-Based Registration and Change Detection

Local Feature Discriminant Projection

Detecting Moving Objects from Dynamic Background Combining Subspace Learning with Mixed Norm Approach

A New Starry Images Matching Method in Dim and Small Space Target Detection

Small and Dim Point Target Detection in Special Feature Space

Learning an Orthogonal and Smooth Subspace for Image Classification.

Local Subspace Collaborative Tracking

MOON: A Subspace-Based Multi-Branch Network for Object Detection in Remotely Sensed Images

Image Classification by Selective Regularized Subspace Learning

On optimal subspaces for appearance-based object recognition

A Novel Approach for Satellite Image Classification Using Local Self-Similarity.

Supervised Local Tangent Space Alignment for Classification

Local Feature Matching Using Deep Learning: A Survey

Multi-view Latent Space Learning Based on Local Discriminant Embedding

Nearest Feature Space Analysis for Classification

Visual Tracking Via Subspace Learning: A Discriminative Approach.

Full-viewpoint 3D Space Object Recognition Based on Kernel Locality Preserving Projections