MTA: A Lightweight Multilingual Text Alignment Model for Cross-Language Visual Word Sense Disambiguation | IEEE Conference Publication | IEEE Xplore