TY  - RPRT
TI  - CaptionFormer: Unified Segmentation, Tracking, and Captioning for Spatio-Temporal Objects
AU  - Gabriel Fiastre
AU  - Antoine Yang
AU  - Cordelia Schmid
PY  - 2026
UR  - https://arxiv.org/abs/2510.14904
ID  - 2510.14904
ER  -