1 paper
Ege Ozguroglu, Ruoshi Liu, Dídac Surís +4
We introduce pix2gestalt, a framework for zero-shot amodal segmentation, which learns to estimate the shape and appearance of whole objects that are only partially visible behind o…