MAF: Multimodal Alignment Framework for Weakly-Supervised Phrase Grounding.

Qinxin Wang Hao Tan Sheng Shen Michael W. Mahoney Zhewei Yao

Published in: EMNLP (1) (2020)

Keyphrases

weakly supervised
multi class
object class
machine learning
computer vision
natural language
training set
semi supervised
unsupervised learning
visual features
topic models
superpixels