A new egocentric benchmark shows vision-language models fail at spatial reasoning across disjoint frames, falling 28 points behind humans and only improving sharply when handed ground-truth 3D coordinates.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Out of Sight, Not Out of Context? Egocentric Spatial Reasoning in VLMs Across Disjoint Frames
A new egocentric benchmark shows vision-language models fail at spatial reasoning across disjoint frames, falling 28 points behind humans and only improving sharply when handed ground-truth 3D coordinates.