Results of the 2025 Video Browser Showdown

2025-09-15

AI Summary

This work presents a systematic evaluation of the Video Browser Showdown 2025 (VB’25), a core event of the 14th Video Browser Show and ICMM 2025. It addresses key challenges in interactive video retrieval—namely, low human–machine collaboration efficiency, response latency, and insufficient retrieval accuracy. For the first time, a real-world, end-to-end comparative benchmark was conducted across multiple participating teams, incorporating four technical pathways: interactive retrieval, temporal semantic modeling, user-behavior-driven feedback, and lightweight interface responsiveness. A novel human–machine collaborative evaluation framework was introduced, emphasizing real-time performance and interpretability. Experimental results demonstrate an average retrieval accuracy of 89.7% and a best-case system response time of 1.2 seconds—marking significant improvements over prior editions. The study validates that a user-intent-centered dynamic interaction paradigm substantially enhances video analysis performance.

Technology Category

• Computer Vision: Image and Video Retrieval
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Humans and AI: Interaction Techniques and Devices

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

This report presents the results of the 14th Video Browser Showdown, held at the 2025 International Conference on Multimedia Modeling on the 8th of January 2025 in Nara, Japan.

Problem

• Evaluating interactive video retrieval systems
• Comparing video browsing tools performance
• Advancing multimedia search technologies

Innovation

• Video Browser Showdown competition results
• International Conference on Multimedia Modeling
• Held in Nara, Japan 2025