Jaehoon Jeong

Temporal Grounding as a Learning Signal for Referring Video Object Segmentation

$\text{O}_\text{2}$VIS: Occupancy-aware Object Association for Temporally Consistent Video Instance Segmentation