CRoPS: A Training-Free Hallucination Mitigation Framework for Vision-Language Models

Neeraj Anand, Samyak Jha, Udbhav Bamba +1

Despite the rapid success of Large Vision-Language Models (LVLMs), a persistent challenge is their tendency to generate hallucinated content, undermining reliability in real-world…