Object detection is a core domain within machine learning. It encounters unique challenges when applied to sports, particularly when it is applied to swing motions –particularly with swiftly moving, small area objects. This research focuses on the specific task of tennis ball detection within frames of professional matches. Tennis balls, with their diminutivesize and speeds exceeding 100 miles per hour, demand precise localization. The methodology involves collecting training and testing data, including over 2000 combined images of generic tennis ball images and frames extracted from professional points sourced by the Association of Tennis Professionals (ATP). Two separate solutions will be studied: Mask R-CNN and YOLOv8. Both of these models aim to address the challenges posed by the distinctive characteristics of tennis balls. While the study pushes questions to be further addressed, its significance lies in contributing insights to the effectiveness of these different models towards high velocity object tracking.