R4DSG builds an anchor-relative 4D scene graph memory from RGB egocentric video and shows improved object-centric QA accuracy over text retrieval.
Marks, and Jonathan Le Roux
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
R4DSG: Relative 4D Scene Graph Memory for Object-Centric Question Answering in Long Egocentric Video
R4DSG builds an anchor-relative 4D scene graph memory from RGB egocentric video and shows improved object-centric QA accuracy over text retrieval.