AI RESEARCH
U-CESE: Unified Clip-based Event Search Engine for AI Challenge HCMC 2025
arXiv CS.CV
•
ArXi:2605.23274v1 Announce Type: new Retrieving events from large-scale video datasets is challenging due to complex temporal, spatial, and multimodal information. This paper presents U-CESE, our solution for the AI Challenge HCMC 2025, a Unified Clip-based Event Search Engine for multimodal event retrieval across diverse video sources. Building on CESE, U-CESE integrates its three modules into a single cohesive framework, ensuring consistent processing and retrieval across query types.