The invention presents a method for segmenting cricket videos by analyzing video frames to detect and map contents, categorizing the video into segments based on content positions, and using color histograms and shape features for further classification. It autonomously identifies and sequences meaningful units, such as 'deliveries', enhancing viewer experience and aiding in efficient video analysis, free from non-game events and advertisements.
In cricket, a ‘delivery’ consists of a bowler’s run-up, followed by the ball being bowled, addressed appropriately by the batsman, and finally fielded and returned. Current methods for segmenting cricket videos rely on manual textual descriptions in commentaries to identify events. Our work automatically segments broadcast video into meaningful entities. This enables a query from a bowling coach such as “Show all deliveries faced by Tendulkar,” returning results in video segments consisting of, say, “the third ball of the first over, the fifth and sixth balls of the second over,” and so on.
- Semi-Supervised Processing: A semi-supervised approach is adopted along with traditional video processing methods.
- Segmentation: It automatically detects and sequences 'deliveries' by segregating 'Field' views into 'Pitch' and 'Ground' views using color and shape features.
- Content Detection: It incorporates a method to capture polygon-shaped boxes with alphanumeric strings for content detection such as scoreboards.
A prototype has been developed.
Results need to be analyzed to validate the process for constantly evolving broadcast videos.
4
This process promotes efficient video analysis, aiding coaches and analysts in player performance evaluation and strategy development. It enhances viewer experience by providing concise, advertisement-free cricket videos focused on game highlights. It also enables storage of compressed videos for archival purposes and deep learning.
- Segmentation of Videos: Video processing platforms can integrate this technology to automate the segmentation of cricket videos, enhancing workflow efficiency and content organization.
- Performance Analysis: Cricket teams and coaches can utilize it for detailed analysis of player performances and opponent strategies.
- Industries/Users: Administrators - ICC, BCCI; TV channels - Star Sports, ESPN; Smartphones - Apple, Samsung, Google; Entertainment Club Owners - IPL family
319/MUM/2015
535031