CORTEXA
← Browse

Ziqi Zhang

2 papers indexed

arxivcs.CV2026-07-14

Gaussian Mixture Modeling for Event-Aware Visual Allocation in Long Video Understanding

Yifan Lu, Ziqi Zhang, Chunfeng Yuan, Jun Gao, Bing Li, Weiming Hu

Large Vision-Language Models (LVLMs) face significant challenges in long video understanding due to the excessive computational cost and information loss associated with uniform sampling. Existing keyframe selection methods often treat video frames as atomic entities and allocate…

View free PDFSource page
arxivcs.CV2026-07-08

MMAgent-R$^2$: Learning to Rerank and Reject for Agentic mRAG

Tao Zhang, Ziqi Zhang, Zongyang Ma, Yuxin Yang, Bing Li, Chunfeng Yuan, et al.

Knowledge-based Visual Question Answering (KB-VQA) requires models to retrieve visual entities matching the query image from large-scale encyclopedic knowledge bases and answer related questions. Existing multimodal Retrieval Augmented Generation (mRAG) methods rely on global vis…

View free PDFSource page