Multi-scale visual attention for attribute disambiguation in zero-shot learning
作者：
Highlights：
• We propose an embedding-based attentional model for zero-shot image recognition.
• Our method combines multi-scale visual attention (VA) and attribute selection (AS).
• The two parts, VA and AS, are optimized in a unified supervised framework.
• Our method can discover the relationships between visual and semantic spaces.
• Experimental results show our method performs better than other related methods.
摘要
•We propose an embedding-based attentional model for zero-shot image recognition.•Our method combines multi-scale visual attention (VA) and attribute selection (AS).•The two parts, VA and AS, are optimized in a unified supervised framework.•Our method can discover the relationships between visual and semantic spaces.•Experimental results show our method performs better than other related methods.
论文关键词：Zero-shot image recognition,Visual attention,Attribute disambiguation
论文评审过程：Received 27 March 2021, Revised 6 November 2021, Accepted 20 December 2021, Available online 4 January 2022, Version of Record 19 January 2022.
论文官网地址：https://doi.org/10.1016/j.image.2021.116614
