SAVE: Sparse Autoencoder-Driven Visual Information Enhancement for Mitigating Object Hallucination.

Sangha Park, Seungryong Yoo, Jisoo Mok, Sungroh Yoon

Browse the full WACV paper archive.