Embedding Visual Hierarchy with Deep Networks for Large-Scale Visual Recognition

TL;DR

This paper introduces a level-wise mixture model that embeds visual hierarchy within deep networks, enabling efficient large-scale visual recognition by jointly learning deep features, hierarchical classifiers, and hierarchy adaptation.

Contribution

It presents a novel end-to-end framework that combines deep networks with hierarchical classification and automatic hierarchy adaptation for large-scale recognition.

Findings

Achieves competitive accuracy on ImageNet datasets.

Supports joint learning for improved recognition performance.

Enhances efficiency in large-scale visual classification.

Abstract

In this paper, a level-wise mixture model (LMM) is developed by embedding visual hierarchy with deep networks to support large-scale visual recognition (i.e., recognizing thousands or even tens of thousands of object classes), and a Bayesian approach is used to adapt a pre-trained visual hierarchy automatically to the improvements of deep features (that are used for image and object class representation) when more representative deep networks are learned along the time. Our LMM model can provide an end-to-end approach for jointly learning: (a) the deep networks to extract more discriminative deep features for image and object class representation; (b) the tree classifier for recognizing large numbers of object classes hierarchically; and (c) the visual hierarchy adaptation for achieving more accurate indexing of large numbers of object classes hierarchically. By supporting joint learning of the tree classifier, the deep networks and the visual hierarchy adaptation, our LMM algorithm can provide an effective approach for controlling inter-level error propagation effectively, thus it can achieve better accuracy rates on large-scale visual recognition. Our experiments are carried on ImageNet1K and ImageNet10K image sets, and our LMM algorithm can achieve very competitive results on both the accuracy rates and the computation efficiency as compared with the baseline methods.