Latent Embedding Feedback and Discriminative Features for Zero-Shot Classification

Zero-shot learning strives to classify unseen categories for which no data is available during training. In the generalized variant, the test samples can further belong to seen or unseen categories. The state-of-the-art relies on Generative Adversarial Networks that synthesize unseen class features by leveraging class-specific semantic embeddings. During training, they generate semantically consistent features, but discard this constraint during feature synthesis and classification. We propose to enforce semantic consistency at all stages of (generalized) zero-shot learning: training, feature synthesis and classification. We first introduce a feedback loop, from a semantic embedding decoder, that iteratively refines the generated features during both the training and feature synthesis stages. The synthesized features together with their corresponding latent embeddings from the decoder are then transformed into discriminative features and utilized during classification to reduce ambiguities among categories. Experiments on (generalized) zero-shot object and action classification reveal the benefit of semantic consistency and iterative feedback, outperforming existing methods on six zero-shot learning benchmarks. Source code at https://github.com/akshitac8/tfvaegan.

PDF Abstract ECCV 2020 PDF ECCV 2020 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Generalized Zero-Shot Learning AwA2 GZSL_TF-VAEGAN Harmonic mean 66.6 # 2
Zero-Shot Learning AwA2 ZSL_TF-VAEGAN average top-1 classification accuracy 72.2 # 2
Zero-Shot Learning CUB-200-2011 ZSL_TF-VAEGAN average top-1 classification accuracy 64.9 # 4
Generalized Zero-Shot Learning CUB-200-2011 GZSL_TF-VAEGAN Harmonic mean 58.1 # 2
Generalized Zero-Shot Learning Oxford 102 Flower GZSL_TF-VAEGAN Harmonic mean 71.7 # 2
Zero-Shot Learning Oxford 102 Flower ZSL_TF-VAEGAN average top-1 classification accuracy 70.8 # 2
Generalized Zero-Shot Learning SUN Attribute GZSL_TF-VAEGAN Harmonic mean 43 # 2
Zero-Shot Learning SUN Attribute ZSL_TF-VAEGAN average top-1 classification accuracy 66 # 2

Methods


No methods listed for this paper. Add relevant methods here