Paper: Midge: Generating Descriptions of Images

ACL ID W12-1523
Title Midge: Generating Descriptions of Images
Venue International Conference on Natural Language Generation
Session Main Conference
Year 2012

We demonstrate a novel, robust vision-to- language generation system called Midge. Midge is a prototype system that connects computer vision to syntactic structures with semantic constraints, allowing for the auto- matic generation of detailed image descrip- tions. We explain how to connect vision de- tections to trees in Penn Treebank syntax, which provides the scaffolding necessary to further refine data-driven statistical generation approaches for a variety of end goals.