CNN with coarse‐to‐fine layer for hierarchical classification. Issue 6 (31st May 2018)

• Record Type: Journal Article
• Authors: Fu, Ruigang; Li, Biao; Gao, Yinghui; Wang, Ping
• Is Part Of: IET computer vision. Volume 12:Issue 6(2018)
• Page Start: 892
• Page End: 899
• Publication Date: 2018-05-31
• Subjects: image classification -- Bayes methods -- stochastic processes -- gradient methods -- neural net architecture; CNN-based classification models -- coarse-to-fine layer -- hierarchical classification -- convolution neural network -- visual separability -- object categories -- category hierarchies -- hierarchical CNN architecture -- Bayesian equation -- stochastic gradient descent -- multiple hierarchical predictions -- benchmark datasets MNIST -- CIFAR-100 dataset -- CIFAR-10 dataset; Computer vision -- Periodicals; Pattern recognition systems -- Periodicals; 006.37
• DOI: 10.1049/iet-cvi.2017.0636
• Languages: English
• ISSNs: 1751-9632

Abstract

Abstract : Most of the traditional convolution neural network (CNN)‐based classification models are flat classifiers, which have an underlying assumption that all classes are equally difficult to distinguish. However, visual separability between different object categories is highly uneven in the real world. Recently, hierarchical classification has been proven effective for CNNs, more and more attempts have been made to exploit category hierarchies in CNN models. In this study, the authors propose a novel hierarchical CNN architecture, called coarse‐to‐fine CNN. It is simple, with a proposed coarse‐to‐fine layer on the top of a generic CNN. The coarse‐to‐fine layer is inspired by the Bayesian equation, where the coarse prediction can affect the fine prediction directly. Arbitrary CNNs can perform the hierarchical classification by adding the proposed layer. The training of a coarse‐to‐fine CNN is end‐to‐end, it can be optimised by typical stochastic gradient descent. In the test phase, it outputs multiple hierarchical predictions simultaneously. Experimental results on the benchmark datasets MNIST, CIFAR‐10, and CIFAR‐100 show clear advantages over the compared baselines.