AI Paper Atlas

Research map / Large language models / Agents, code and math

VIRO: Robust and Efficient Neuro-Symbolic Reasoning with Verification for Referring Expression Comprehension

Hyejin Park, Junhyuk Kwon, Suha Kwak, Jungseul Ok

CVPR 2026 · Large language models › Agents, code and math · approach: Inductive and abstract reasoning evaluation · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add operator-based programs (no verification) to detector-only (ablation)GroundingDINO-T · gRefCOCO no-target + RefCOCO TestAbalanced accuracy: 40.0 -> 56.8Improved
add logical verification (LV) to operators (ablation)VIRO · gRefCOCO no-target + RefCOCO TestAbalanced accuracy: 56.8 -> 57.0No clear effect
add uncertainty verification with fixed threshold (ablation)VIRO · gRefCOCO no-target + RefCOCO TestAbalanced accuracy: 57.0 -> 58.8Improved
use adaptive per-label threshold instead of fixed in uncertainty verification (ablation)VIRO · gRefCOCO no-target + RefCOCO TestAbalanced accuracy: 58.8 -> 61.1Improved
add verification (LV and UV adaptive) to operators (ablation)VIRO · RefCOCO testATPR (Acc@0.5): 74.6 -> 71.9Hurt
add uncertainty verification (adaptive) to operators (ablation)VIRO · gRefCOCO no-target (TestA)TNR (N-acc): 39.3 -> 50.2Improved
use CLIP ViT-H/14 instead of ViT-L/14 (ablation)VIRO · RefCOCO testATPR (Acc@0.5): 68.8 -> 71.9Improved
raise the OVD detection threshold (ablation)VIRO · RefCOCO testA / gRefCOCOTPR and TNRMixed
replace Python-code generation and iterative planners with decoupled symbolic programs and validatorVIRO vs HYDRA · RefCOCOprogram failure rate (%): 28.46 -> 0.07Improved
decouple program generation from execution (amortize one program over N images)VIRO · 1-query-N-imagestotal inference timeImproved
VIRO vs ViperGPT end-to-end latency (validator loop adds LLM cost)VIRO · RefCOCO testAE2E runtime (s/query): 30.19 -> 12.92Improved
VIRO vs NAVER on RefCOCOg exclusive accuracyVIRO · RefCOCOg testAcc@0.5 (exclusive): 68.4 -> 66.6Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy