Attribute annotation on large-scale image database by active knowledge transfer

作者：

Highlights：

• Combining transfer learning and active learning will reduce the annotation workload.

• Attribute relationships are helpful in the attribute learning process.

• Relationships between attributes and categories are learned on the annotated images.

摘要

•Combining transfer learning and active learning will reduce the annotation workload.•Attribute relationships are helpful in the attribute learning process.•Relationships between attributes and categories are learned on the annotated images.

论文关键词：Attribute,Annotation,Relationship,Active learning,Transfer learning

论文评审过程：Received 5 April 2017, Revised 7 May 2018, Accepted 29 June 2018, Available online 20 July 2018, Version of Record 7 August 2018.

论文官网地址：https://doi.org/10.1016/j.imavis.2018.06.012