What Makes "Good" Distractors for Object Hallucination Evaluation in Large Vision-Language Models?

Open in new window