An Empirical Study and Analysis of Generalized Zero-Shot Learning for Object Recognition in the Wild

Abstract

Zero-shot learning (ZSL) methods have been studied in the unrealistic setting where test data are assumed to come from unseen classes only. In this paper, we advocate studying the problem of generalized zero-shot learning (GZSL) where the test data's class memberships are unconstrained. We show empirically that naively using the classifiers constructed by ZSL approaches does not perform well in the generalized setting. Motivated by this, we propose a simple but effective calibration method that can be used to balance two conflicting forces: recognizing data from seen classes versus those from unseen ones. We develop a performance metric to characterize such a trade-off and examine the utility of this metric in evaluating various ZSL approaches. Our analysis further shows that there is a large gap between the performance of existing approaches and an upper bound established via idealized semantic embeddings, suggesting that improving class semantic embeddings is vital to GZSL.

ECCV2016 camera-ready

References in corpus (2)

• Transductive Multi-view Zero-Shot Learning
• Synthesized Classifiers for Zero-Shot Learning

Cited by in corpus (40)

• Semantic Autoencoder for Zero-Shot Learning
• Zero-Shot Learning -- The Good, the Bad and the Ugly
• A causal view of compositional zero-shot recognition
• Low-Shot Learning from Imaginary Data
• From Zero-shot Learning to Conventional Supervised Classification: Unseen Visual Data Synthesis
• Zero-shot Recognition via Semantic Embeddings and Knowledge Graphs
• Transductive Unbiased Embedding for Zero-Shot Learning
• Discriminative Learning of Latent Features for Zero-Shot Recognition
• Multi-Head Self-Attention via Vision Transformer for Zero-Shot Learning
• Mitigating the Hubness Problem for Zero-Shot Learning of 3D Objects
• Zero-Shot Visual Recognition using Semantics-Preserving Adversarial Embedding Networks
• Learning Class Prototypes via Structure Alignment for Zero-Shot Recognition
• Learning Joint Feature Adaptation for Zero-Shot Recognition
• Relational Generalized Few-Shot Learning
• Predicting Visual Exemplars of Unseen Classes for Zero-Shot Learning
• A Meta-Learning Framework for Generalized Zero-Shot Learning
• Combinets: Creativity via Recombination of Neural Networks
• Zero-shot Learning via Shared-Reconstruction-Graph Pursuit
• Zero-Shot Semantic Segmentation
• Attribute Prototype Network for Zero-Shot Learning
• Adaptive Confidence Smoothing for Generalized Zero-Shot Learning
• Region Semantically Aligned Network for Zero-Shot Learning
• Deep Cross-Modal Audio-Visual Generation
• Generalized Zero-shot ICD Coding
• Domain segmentation and adjustment for generalized zero-shot learning
• Zero and Few Shot Learning with Semantic Feature Synthesis and Competitive Learning
• Classifier and Exemplar Synthesis for Zero-Shot Learning
• Zero-Shot Learning via Category-Specific Visual-Semantic Mapping
• Bi-Adversarial Auto-Encoder for Zero-Shot Learning
• Alternative Semantic Representations for Zero-Shot Human Action Recognition
• Generalized Zero-Shot Learning for Action Recognition with Web-Scale Video Data
• Global Semantic Consistency for Zero-Shot Learning
• Infinite-Label Learning with Semantic Output Codes
• Learning Classifiers for Domain Adaptation, Zero and Few-Shot Recognition Based on Learning Latent Semantic Parts
• Bias-Awareness for Zero-Shot Learning the Seen and Unseen
• Fine-grained Classification using Heterogeneous Web Data and Auxiliary Categories
• Generalised Zero-Shot Learning with a Classifier Ensemble over Multi-Modal Embedding Spaces
• Beyond Attributes: Adversarial Erasing Embedding Network for Zero-shot Learning
• Exploring the Challenges towards Lifelong Fact Learning
• Learning Single/Multi-Attribute of Object with Symmetry and Group