Yu-Feng Yen

Attribution Gaps in Zero-Training LLM+OVOD Pipelines: A Fine-Grained Analysis of the CAAP--SNAP Discrepancy

The upshot is that a large share of the apparent CAAP--SNAP gap traces back to closed-category annotation limits rather than a real grounding failure, which matters for how the authors detect hallucination, analyze failure modes, and design evaluation for grounded multimodal systems meant to work in the open world.