Optimization Research on Deep Learning and Temporal Segmentation Algorithm of Video Shot in Basketball Games

Area of Science:

• Computer Vision
• Multimedia Analysis
• Sports Analytics

Background:

• Temporal segmentation of video shots is crucial for video applications.
• Current methods for basketball game video segmentation are time-consuming.
• Optimizing video analysis in sports requires efficient algorithms.

Purpose of the Study:

• To propose a novel deep learning model for temporal segmentation of basketball game videos.
• To develop a histogram-based algorithm to address the long segmentation time of existing methods.
• To enhance the efficiency and accuracy of video shot analysis in sports.

Main Methods:

• Utilized deep learning for boundary detection and frame processing.
• Converted video data from RGB to HSV color space.
• Applied histogram statistics for dimensionality reduction and feature vector creation.
• Calculated frame differences and other metrics, combined with a dynamic threshold for segmentation optimization.

Main Results:

• The proposed algorithm demonstrates improved efficiency in temporal segmentation of basketball game videos.
• Effectiveness verified through reduced missed detection rates.
• The optimized algorithm achieved efficient implementation in terms of split time.

Conclusions:

• The developed deep learning and histogram-based approach offers an efficient solution for basketball video shot temporal segmentation.
• This method significantly improves upon existing algorithms in terms of speed and accuracy.
• The findings have practical significance and application prospects in multimedia research and sports analytics.