Abstract:Zero-shot image recognition attempts to simulate the zero-shot learning mechanism of humans and recognizes the images of novel classes. It is crucial to learn transferable knowledge from seen classes and generalize it to unseen classes for image recognition in zero-shot learning (ZSL). Most existing ZSL methods extract visual features with pretrained backbone networks and learn transferable knowledge with the extracted visual features. However, the backbone networks are not pretrained for a special task, and the extracted visual features usually contain some distractive information for the ZSL task, which causes some discriminative information to be ignored or weakened and degrades the quality of knowledge learned from seen classes. Moreover, since visual samples of unseen classes are not obtainable, domain shift is another challenging problem. In this paper, we propose visual feature enhancement to learn more discriminative visual features via a graph convolutional network (GCN) and an attention mechanism for improving the quality of the learned transferable knowledge. Different from previous works, we explore the correlations between different latent visual patterns of an image and introduce GCN to enhance visual features. On the other hand, we take advantage of different learning mechanisms of GCN and MLP and propose dual classifier learning for improving the generalization and inference capabilities of our model. In end-to-end model training, the module of visual feature enhancement and the module of dual classifier learning are beneficial to each other via joint optimization. Finally, we perform extensive experiments in the ZSL setting and GZSL setting. The extensive experimental results verify the effectiveness and superiority of our method.

Class-Incremental Generalized Zero-Shot Learning.

Zero-Shot Learning with Generative Latent Prototype Model.

Incremental Zero-Shot Learning

Transfer Increment for Generalized Zero-Shot Learning

Cluster-based Contrastive Disentangling for Generalized Zero-Shot Learning

MetaZSCIL: A Meta-Learning Approach for Generalized Zero-Shot Class Incremental Learning.

Evolutionary Generalized Zero-Shot Learning

Learn More from Less: Generalized Zero-Shot Learning with Severely Limited Labeled Data

OntoZSL: Ontology-enhanced Zero-shot Learning

Learning Discriminative Domain-Invariant Prototypes for Generalized Zero Shot Learning

Lifelong Zero-Shot Learning.

Learning Discriminative Latent Features for Generalized Zero-and Few-Shot Learning.

Learning Multiple Criteria Calibration for Generalized Zero-shot Learning

Zero-shot Learning Via the Fusion of Generation and Embedding for Image Recognition

Improving generalized zero-shot learning via cluster-based semantic disentangling representation

A Review of Generalized Zero-Shot Learning Methods

Prototype-Augmented Self-Supervised Generative Network for Generalized Zero-Shot Learning.

Zero-shot learning via visual feature enhancement and dual classifier learning for image recognition

Less but Better: Enabling Generalized Zero-shot Learning Towards Unseen Domains by Intrinsic Learning from Redundant LLM Semantics

Domain-aware Prototype Network for Generalized Zero-Shot Learning

One-Stage Training Generative Paradigm for Generalized Zero-Shot Learning.