A combined rule-based plus LLM-judge step reward, optimized with GRPO, improves multi-step visual tool orchestration for video anomaly detection on the new VAD-Tool benchmark.
Title resolution pending
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
-
VTO: Visual Tool Orchestration for Video Anomaly Detection
A combined rule-based plus LLM-judge step reward, optimized with GRPO, improves multi-step visual tool orchestration for video anomaly detection on the new VAD-Tool benchmark.