Paper: Generalizing Image Captions for Image-Text Parallel Corpus

ACL ID P13-2138
Title Generalizing Image Captions for Image-Text Parallel Corpus
Venue Annual Meeting of the Association of Computational Linguistics
Session Short Paper
Year 2013

The ever growing amount of web images and their associated texts offers new op- portunities for integrative models bridging natural language processing and computer vision. However, the potential benefits of such data are yet to be fully realized due to the complexity and noise in the align- ment between image content and text. We address this challenge with contributions in two folds: first, we introduce the new task of image caption generalization, for- mulated as visually-guided sentence com- pression, and present an efficient algo- rithm based on dynamic beam search with dependency-based constraints. Second, we release a new large-scale corpus with 1 million image-caption pairs achieving tighter content alignment between images and text. Evaluation results show the in- trinsic quality ...