Reproduce “Shot Segmentation Based on Von Neumann Entropy for Key Frame Extraction” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xueqing Zhang, Di Fu, Naihao Liu
Researchers from Xi’an Jiaotong University and ByteDance introduce a method for video key frame extraction that segments videos using Von Neumann entropy. This approach accurately identifies representative frames while minimizing redundancy, outperforming a Density Peak Clustering baseline with higher effective information rates and lower redundancy rates.
Key insights
- Von Neumann entropy, adapted from quantum information theory, can effectively measure content change between video frames, enabling adaptive shot segmentation.
- Minimizing the sum of entropies for segmented shots provides a robust criterion for automatically determining optimal shot boundaries and, consequently, the number of key frames.
- Selecting the first frame of each segmented shot naturally incorporates temporal information, ensuring that key frames represent distinct content transitions and reducing redundancy.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.