Addressing object hallucination in large vision-language models through a post-hoc rectification approach.
