Vision-language models can produce crash diagrams from reports with moderate quality, GPT-4o scoring highest at 6.29/10 across 79 roundabout cases using a structured prompt and 10-metric evaluation.
arXiv preprint arXiv:2507.02074 (2025)
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
TrafficRAG is a multimodal RAG framework that turns accident videos into structured text queries, retrieves legal rules and cases via BM25 plus dense retrieval, and generates liability reports, reporting 77.32% legal norm accuracy and 5.48% liability ratio MAE.
citing papers explorer
-
Automating Crash Diagram Generation Using Vision-Language Models: A Case Study on Multi-Lane Roundabouts
Vision-language models can produce crash diagrams from reports with moderate quality, GPT-4o scoring highest at 6.29/10 across 79 roundabout cases using a structured prompt and 10-metric evaluation.
-
TrafficRAG: A Multimodal RAG Framework for Traffic Accident Liability Determination
TrafficRAG is a multimodal RAG framework that turns accident videos into structured text queries, retrieves legal rules and cases via BM25 plus dense retrieval, and generates liability reports, reporting 77.32% legal norm accuracy and 5.48% liability ratio MAE.