Spherical Feature Transform for Deep Metric Learning

by   Yuke Zhu, et al.
Megvii Technology Limited
Tongji University

Data augmentation in feature space is effective to increase data diversity. Previous methods assume that different classes have the same covariance in their feature distributions. Thus, feature transform between different classes is performed via translation. However, this approach is no longer valid for recent deep metric learning scenarios, where feature normalization is widely adopted and all features lie on a hypersphere. This work proposes a novel spherical feature transform approach. It relaxes the assumption of identical covariance between classes to an assumption of similar covariances of different classes on a hypersphere. Consequently, the feature transform is performed by a rotation that respects the spherical data distributions. We provide a simple and effective training method, and in depth analysis on the relation between the two different transforms. Comprehensive experiments on various deep metric learning benchmarks and different baselines verify that our method achieves consistent performance improvement and state-of-the-art results.


page 1

page 2

page 3

page 4


Online Deep Metric Learning

Metric learning learns a metric function from training data to calculate...

Coded Residual Transform for Generalizable Deep Metric Learning

A fundamental challenge in deep metric learning is the generalization ca...

Sharing Matters for Generalization in Deep Metric Learning

Learning the similarity between images constitutes the foundation for nu...

Learning with Memory-based Virtual Classes for Deep Metric Learning

The core of deep metric learning (DML) involves learning visual similari...

Implicit Semantic Augmentation for Distance Metric Learning in Domain Generalization

Domain generalization (DG) aims to learn a model on one or more differen...

Learning from Between-class Examples for Deep Sound Recognition

Deep learning methods have achieved high performance in sound recognitio...

Invariant Feature Mappings for Generalizing Affordance Understanding Using Regularized Metric Learning

This paper presents an approach for learning invariant features for obje...

Please sign up or login with your details

Forgot password? Click here to reset