Learning the Redundancy-free Features for Generalized Zero-Shot Object Recognition
Abstract
Zero-shot object recognition or zero-shot learning aims to transfer the object recognition ability among the semantically related categories, such as fine-grained animal or bird species. However, the images of different fine-grained objects tend to merely exhibit subtle differences in appearance, which will severely deteriorate zero-shot object recognition. To reduce the superfluous information in the fine-grained objects, in this paper, we propose to learn the redundancy-free features for generalized zero-shot learning. We achieve our motivation by projecting the original visual features into a new (redundancy-free) feature space and then restricting the statistical dependence between these two feature spaces. Furthermore, we require the projected features to keep and even strengthen the category relationship in the redundancy-free feature space. In this way, we can remove the redundant information from the visual features without losing the discriminative information. We extensively evaluate the performance on four benchmark datasets. The results show that our redundancy-free feature based generalized zero-shot learning (RFF-GZSL) approach can achieve competitive results compared with the state-of-the-arts.
Cite
@article{arxiv.2006.08939,
title = {Learning the Redundancy-free Features for Generalized Zero-Shot Object Recognition},
author = {Zongyan Han and Zhenyong Fu and Jian Yang},
journal= {arXiv preprint arXiv:2006.08939},
year = {2021}
}
Comments
Some researchers and we have found KNN results in 1st version are incorrect, due to a careless mistake in the code. Concretely, the parameters for accuracy function of KNN were organized in the wrong order by mistake. The softmax results are correct. We have removed all KNN results and remove the SOTA claims. According to the Program Chairs' suggestion, we have made errata request to CVF and IEEE