E2LVLM improves out-of-context misinformation detection by having a vision-language model rerank and rewrite retrieved evidence, then fine-tune on LVLM-generated explanations.
Overview of the grand challenge on detect- ing cheapfakes at acm icmr 2024
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
E2LVLM:Evidence-Enhanced Large Vision-Language Model for Multimodal Out-of-Context Misinformation Detection
E2LVLM improves out-of-context misinformation detection by having a vision-language model rerank and rewrite retrieved evidence, then fine-tune on LVLM-generated explanations.