
Appearance information enhances object association accuracy in video instance segmentation.


coremsg

visage-video-instance-segmentation-with-appearance-guided-enhancement


VISAGE: Video Instance Segmentation with Appearance-Guided Enhancement



Appearance information enhances object association in video instance segmentation.



Video Instance Segmentation with Appearance-Guided Enhancement: VISAGE Study



Providing taxonomy information enhances video instance segmentation performance across multiple datasets.



Taxonomy-aware Multi-dataset Joint Training for Video Instance Segmentation
