v
We invite researchers and practitioners in computer vision, information retrieval, and human-computer interaction to participate in the 1st International Workshop on Interactive Video Search and Exploration (IViSE 2025), held in conjunction with CVPR 2025. Date: June 11 - June 15 Location: CVPR 2025 @ Nashville TN Website: https://sites.google.com/view/ivise2025 Workshop Overview The field of video understanding and retrieval has made significant strides in recent years, but current AI-driven methods struggle with long-form video content. IViSE 2025 aims to address this challenge by exploring human-machine teaming approaches for video search and question answering (QA). This workshop will serve as a venue to compare "fully automated" approaches with "interactive, human-in-the-loop" systems for video retrieval. The focus will be on real-world scenarios involving large-scale video archives, bridging gaps between machine learning, computer vision, and user-centered interactive search strategies. IViSE 2025 Challenge: Text-Based Video Retrieval and Question Answering The IViSE 2025 challenge will use the Vimeo Creative Commons Collection (V3C) dataset—comprising 7,475 videos with a total duration of 1,000 hours—to evaluate state-of-the-art methods in: 1 Fully-Automated Track Queries are provided in advance. Participants develop end-to-end AI-based video retrieval and QA systems. Solutions are evaluated on accuracy and retrieval effectiveness. 2 Interactive Track Queries are presented during the workshop. Teams must solve tasks in real-time within five minutes, leveraging human-machine collaboration. Inspired by the Video Browser Showdown and Lifelog Search Challenge: Known-Item Search (KIS): Given a textual description, retrieve the correct video segment and time interval. Question Answering (QA): Answer a question based on a textual description of a video. Both tracks will use standardized evaluation metrics and build upon existing benchmarks such as TRECVID & DVU for automated search, and Video Browser Showdown for interactive retrieval. Why Participate? Advance the field of long-form video retrieval and exploration. Benchmark your methods against cutting-edge AI and interactive systems. Engage with leading researchers in computer vision, video retrieval, and interactive search. Compete in a structured challenge format with an emphasis on both fully automated & interactive approaches and also in a high-impact challenge at CVPR 2025. Important Dates Feb 3, 2025 - Fully-Automated Track queries released Mar 24, 2025 - Submission deadline for Fully-Automated Track results & workshop papers Apr 3, 2025 - Reviews and challenge scores released Apr 7, 2025 - Camera-ready deadline We look forward to your participation in IViSE 2025 as we explore the next frontier in interactive video search and long-form video understanding