notesum.ai

Published at December 10

cs.CV

Released Date: December 10, 2024

Authors: Yicheng Wang¹, Zhikang Zhang², Jue Wang², David Fan², Zhenlin Xu², Linda Liu², Xiang Hao², Vimal Bhat², Xinyu Li²

Aff.: ¹Texas A&M University; ²Amazon

\hlineB3 Method	Relation	Speaking	Scene
VideoBERT [50]	52.4	37.9	54.9
Obj.T4mer [62]	53.1	39.4	56.9
LST[23]	52.4	37.3	62.8
Orthoformer [45]	50.0	38.3	66.3
ViS4mer[23]	57.1	40.8	67.4
Ours	61.9	42.7	70.9
\rowcolorgray!10 LF-VILA[51]	61.5	41.3	68.0
\rowcolorgray!10S5 ${}_{\text{60frame}}$ [57]	61.9	41.8	69.9
\rowcolorgray!10 S5 ${}_{\text{100frame}}$ [57]	66.7	41.8	73.3
\rowcolorgray!10 MA-LMM ${}_{\text{100frame}}$ [22]	58.2	44.8	80.3
\hlineB3