While there has been significant recent work on learning semantic parsers for specific task/ domains, the results don’t transfer from one domain to another domains. We describe a project to learn a broad-coverage semantic lexicon for domain independent semantic parsing. The technique involves several bootstrapping steps starting from a semantic parser based on a modest-sized hand-built semantic...

AbstractIn this paper, we study the task of synthetic-to-real domain generalized semantic segmentation, which aims to learn a model that is robust unseen real-world scenes using only synthetic data. The large shift between and data, including limited source environmental variations distribution gap significantly hinders performance on scenes. In work, propose Style-HAllucinated Dual consistEncy...

Training a deep neural model for semantic segmentation requires collecting large amount of pixel-level labeled data. To alleviate the data scarcity problem presented in real world, one could utilize synthetic whose label is easy to obtain. Previous work has shown that performance can be improved by training jointly with and examples proper weighting on Such was learned heuristic maximize simila...