ReXSonoVQA is a new video QA benchmark for procedure-centric ultrasound that shows VLMs extract some information but struggle with troubleshooting and causal reasoning compared to text-only baselines.
When the operator sweeps inferiorly and identifies the bifurcation, what anatomical landmark is being visualized?
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
ReXSonoVQA: A Video QA Benchmark for Procedure-Centric Ultrasound Understanding
ReXSonoVQA is a new video QA benchmark for procedure-centric ultrasound that shows VLMs extract some information but struggle with troubleshooting and causal reasoning compared to text-only baselines.