Grounding Agentic VLMs with Dedicated Segmentation for Fine-Grained Vehicle Damage Assessment

Vishwajeet Shivaji Hogale, Anjali Pai, Nitya Ravi

Vision-language models (VLMs) are increasingly deployed as reasoning agents in real-world visual assessment pipelines, yet their spatial grounding remains unreliable for fine-grain…