AI RESEARCH
SGTA: Scene-Graph Based Multi-Modal Traffic Agent for Video Understanding
arXiv CS.CV
•
ArXi:2604.03697v1 Announce Type: new We present Scene-Graph Based Multi-Modal Traffic Agent (SGTA), a modular framework for traffic video understanding that combines structured scene graphs with multi-modal reasoning. It constructs a traffic scene graph from roadside videos using detection, tracking, and lane extraction, followed by tool-based reasoning over both symbolic graph queries and visual inputs. SGTA adopts ReAct to process interleaved reasoning traces from large language models with tool invocations, enabling interpretable decision-making for complex video questions.