Texture Classification in Extreme Scale Variations using GANet

TL;DR

This paper introduces a novel approach combining a GANet with a Genetic Algorithm and a Fisher Vector CNN for classifying textures under extreme scale variations, supported by a new dataset, ESVaT.

Contribution

It presents the first study on extreme scale variation in texture classification, proposing a new GANet architecture and a dedicated dataset for this challenging problem.

Findings

Outperforms standard texture features by over 10% on ESVaT.

Demonstrates superior accuracy on KTHTIPS2b, OS, and synthetic Forrest datasets.

Validates the effectiveness of the proposed framework for extreme scale variations.

Abstract

Research in texture recognition often concentrates on recognizing textures with intraclass variations such as illumination, rotation, viewpoint and small scale changes. In contrast, in real-world applications a change in scale can have a dramatic impact on texture appearance, to the point of changing completely from one texture category to another. As a result, texture variations due to changes in scale are amongst the hardest to handle. In this work we conduct the first study of classifying textures with extreme variations in scale. To address this issue, we first propose and then reduce scale proposals on the basis of dominant texture patterns. Motivated by the challenges posed by this problem, we propose a new GANet network where we use a Genetic Algorithm to change the units in the hidden layers during network training, in order to promote the learning of more informative semantic texture patterns. Finally, we adopt a FVCNN (Fisher Vector pooling of a Convolutional Neural Network filter bank) feature encoder for global texture representation. Because extreme scale variations are not necessarily present in most standard texture databases, to support the proposed extreme-scale aspects of texture understanding we are developing a new dataset, the Extreme Scale Variation Textures (ESVaT), to test the performance of our framework. It is demonstrated that the proposed framework significantly outperforms gold-standard texture features by more than 10% on ESVaT. We also test the performance of our proposed approach on the KTHTIPS2b and OS datasets and a further dataset synthetically derived from Forrest, showing superior performance compared to the state of the art.