Multimodal Large Language Models for Visual Attribute Inference in iRAP Road Safety Assessment | Publicación