Search papers, labs, and topics across Lattice.
CASIA, Key Laboratory of Super Intelligent Security of Multi-Modal Information
2
0
4
Visual reranking and active rejection in MMAgent-R$^2$ significantly boost retrieval accuracy in challenging KB-VQA tasks, outperforming traditional methods.
MLLMs can be tricked into missing 90% of harmful content simply by encoding it in images that humans can easily read.
