Zero-Shot Classification by Generating Artificial Visual Features
Résumé
This paper addresses the task of learning an image clas-sifier when some categories are defined by semantic descriptions only (visual attributes) while the others are also defined by exemplar images. This task is often referred to as Zero-Shot classification (ZSC). Most of the previous methods rely on learning a common embedding space allowing to compare visual features of unknown categories with semantic descriptions. This paper argues that these approaches are limited as i) efficient discriminative classi-fiers can't be used and ii) classification tasks with seen and unseen categories (Generalized Zero-Shot Classification or GZSC) can't be addressed efficiently. This paper suggests to address ZSC and GZSC by i) learning a conditional generator using seen classes ii) generate artificial training examples for the categories without exemplars. ZSC is therefore turned into a standard supervised learning problem. Experiments with 4 generative models and 6 datasets experimentally validate the approach, giving state-of-the-art results on both ZSC and GZSC.
Origine : Fichiers produits par l'(les) auteur(s)