Recognizing open-vocabulary relations between objects in images

How can we represent relations between objects in images using languare? Recently, deep neural networks achieved significant performances on various object recognition tasks. We focuse on relations of such objects and propose a model for representing the relations by language. This project is a collaboration with Communication Science Laboratory at Tohoku University.


Recognizing Open-Vocabulary Relations between Objects in Images
Pacific Asia Conference on Language, Information and Computation (PACLIC) 2016
Muraoka, M., Maharjan, S., Saito, M., Yamaguchi, K., Okazaki, N., Okatani, T., & Inui, K.