Atlantis: Aesthetic-oriented multiple granularities fusion network for joint multimodal aspect-based sentiment analysis

Luwei Xiao, Xingjiao Wu, Junjie Xu, Weijie Li, Cheng Jin, Liang He

Abstract: Highlights•Atlantis enables an efficient cross-modal alignment and multiple granularities fusion.•We are the first to incorporate image aesthetic assessment for JMASA.•We explore the relationship between image attributes and multimodal sentiment analysis.•Our model achieves state-of-the-art results.