AI-Based Cropping of Soccer Videos for Different Social Media Representations
Houshmand Sarkhoosh, Mehdi; Dorcheh, Sayed Mohammad Majidi; Midoglu, Cise; Sabet, Saeed; Kupka, Tomas; Johansen, Dag; Riegler, Michael Alexander; Halvorsen, Pål
Chapter, Peer reviewed, Conference object, Journal article
Accepted version
View/ Open
Date
2024Metadata
Show full item recordCollections
Original version
http://dx.doi.org/10.1007/978-3-031-53302-0_22Abstract
The process of re-publishing soccer videos on social media often involves labor-intensive and tedious manual adjustments, particularly when altering aspect ratios while trying to maintain key visual elements. To address this issue, we have developed an AI-based automated cropping tool called SmartCrop which uses object detection, scene detection, outlier detection, and interpolation. This innovative tool is designed to identify and track important objects within the video, such as the soccer ball, and adjusts for any tracking loss. It dynamically calculates the cropping center, ensuring the most relevant parts of the video remain in the frame. Our initial assessments have shown that the tool is not only practical and efficient but also enhances accuracy in maintaining the essence of the original content. A user study confirms that our automated cropping approach significantly improves user experience compared to static methods. We aim to demonstrate the full functionality of SmartCrop, including visual output and processing times, highlighting its efficiency, support of various configurations, and effectiveness in preserving the integrity of soccer content during aspect ratio adjustments.