Hierarchical Metric Learning for Optical Remote Sensing Scene Categorization

TL;DR

This paper introduces a hierarchical metric learning approach for optical remote sensing scene classification, organizing classes based on visual similarities and learning separate metrics to improve fine-grained recognition accuracy.

Contribution

It proposes a novel hierarchical organization of classes and learns class-specific distance metrics, addressing limitations of standard metric learning methods.

Findings

Improved classification accuracy on NWPU-RESISC45 dataset

Effective hierarchical class organization based on visual similarities

Outperforms standard metric learning approaches

Abstract

We address the problem of scene classification from optical remote sensing (RS) images based on the paradigm of hierarchical metric learning. Ideally, supervised metric learning strategies learn a projection from a set of training data points so as to minimize intra-class variance while maximizing inter-class separability to the class label space. However, standard metric learning techniques do not incorporate the class interaction information in learning the transformation matrix, which is often considered to be a bottleneck while dealing with fine-grained visual categories. As a remedy, we propose to organize the classes in a hierarchical fashion by exploring their visual similarities and subsequently learn separate distance metric transformations for the classes present at the non-leaf nodes of the tree. We employ an iterative max-margin clustering strategy to obtain the hierarchical organization of the classes. Experiment results obtained on the large-scale NWPU-RESISC45 and the popular UC-Merced datasets demonstrate the efficacy of the proposed hierarchical metric learning based RS scene recognition strategy in comparison to the standard approaches.