1 paper · 1 filter
Savindu Dilshan Wickramasinghe
Referring expression segmentation requires language conditioned localization and pixel-accurate masks, but monolithic models can be costly to deploy. We present VespaSeg, a modular…