Scholarlink

<p>The goal of this project is to develop a video summarizing system that can produce an informative synopsis of the source video. A novel video summarization system is proposed for employing optimized data-cube search for shot boundary detection called as VSUMMDCube (Video Summarization using Datacube Search). By dividing videos into segments using datacube search, the method accurately identifies shot boundaries and extracts key frames based on visual features like color, edge, and texture. Comparative analysis with state-of-the-art techniques demonstrates superior performance, achieving the highest F1 scores of 51.7% and 60.2% on the TVSum and SUMMe datasets, respectively. The code and datasets used in the manuscript can be accessed through <a href="https://github.com/adstechlearning/VideoAnalysis">https://github.com/adstechlearning/VideoAnalysis</a>. The proposed method enhances compression ratio while preserving crucial video content, offering a practical solution for managing large video datasets. In final analysis, the empirical results of the proposed approach and the current methodologies are presented in this research using metrics that determine the accuracy of the work and demonstrate strong practical significance.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Optimized data-cube search for enhanced video summarization via shot boundary detection

  • 期刊论文

  • 发表日期：

    2026年5月29日

查看全文
  • J. Kavitha,
  • S. Alex David,
  • M. Mohamed Iqbal,
  • Anil Kumar Bisht,
  • Ankur Dumka,
  • Rajesh Singh,
  • Anita Gehlot,
  • Amit Kumar Thakur,
  • Sudhanshu Dogra

摘要

The goal of this project is to develop a video summarizing system that can produce an informative synopsis of the source video. A novel video summarization system is proposed for employing optimized data-cube search for shot boundary detection called as VSUMMDCube (Video Summarization using Datacube Search). By dividing videos into segments using datacube search, the method accurately identifies shot boundaries and extracts key frames based on visual features like color, edge, and texture. Comparative analysis with state-of-the-art techniques demonstrates superior performance, achieving the highest F1 scores of 51.7% and 60.2% on the TVSum and SUMMe datasets, respectively. The code and datasets used in the manuscript can be accessed through https://github.com/adstechlearning/VideoAnalysis. The proposed method enhances compression ratio while preserving crucial video content, offering a practical solution for managing large video datasets. In final analysis, the empirical results of the proposed approach and the current methodologies are presented in this research using metrics that determine the accuracy of the work and demonstrate strong practical significance.