CARVE compresses 3D medical volume tokens by adaptively windowing the depth axis, selecting anchors and retrieved evidence, and folding redundant tokens, preserving near-full VQA quality at about one fifth of the tokens.
Advances in Neural Information Processing Systems , volume=
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
-
CARVE: Cross-Slice Anisotropic Reallocation of Visual Evidence for Efficient 3D Medical Volume Understanding
CARVE compresses 3D medical volume tokens by adaptively windowing the depth axis, selecting anchors and retrieved evidence, and folding redundant tokens, preserving near-full VQA quality at about one fifth of the tokens.