Related Experiment Videos
Two-Way Affective Modeling for Hidden Movie Highlights' Extraction
Zheng Wang1, Xinyu Yan2, Wei Jiang3,4
1Division of Intelligence and Computing, Tianjin University, Tianjin 300072, China. wzheng@tju.edu.cn.
Sensors (Basel, Switzerland)
|December 6, 2018
Summary
This study introduces a novel two-way excitement model for automatic movie highlight extraction. It effectively identifies hidden highlights by analyzing middle-level features, improving content analysis and trailer production.
Area of Science:
- Computer Science
- Multimedia Analysis
- Artificial Intelligence
Background:
- Automatic movie highlight extraction is crucial for content analysis and production.
- Previous methods often miss highlights with low-level audio-visual features.
- Cognitive psychology principles can inform better highlight detection.
Purpose of the Study:
- To develop an improved method for automatic movie highlight extraction.
- To address the limitation of previous approaches that ignore "hidden" highlights.
- To bridge the gap between low-level features and high-level perceptual excitement.
Main Methods:
- A two-way excitement model combining top-down and bottom-up processing.
- Extraction of middle-level features based on global sensitivity and local abnormality.
- Validation using a diverse set of well-known movies.
Main Results:
- The proposed model successfully extracts movie highlights, including those with low-level feature responses.
- Quantitative assessments confirm the method's effectiveness.
- The approach demonstrates promising results in identifying audience excitement.
Conclusions:
- The novel two-way excitement model enhances automatic movie highlight extraction.
- Analyzing middle-level features is key to capturing subtle highlights.
- This method offers significant improvements for content analysis, ranking, indexing, and trailer production.