Ball-Type Small Objection Algorithm Based on YOLOv8

Object detection is an important computer vision task in digital images for a specific class of phenomena (such as people, animals or ball-type objects). To improve the accuracy of small object detection for high-speed moving balls in sports fields, we researched and tested a small object detection model based on the YOLOv8 algorithm. The YOLO algorithm based on deep learning is fast and accurate in operation, making it suitable for real-time systems. On the basis of the original algorithm, we carried out algorithm fusion and parameter adjustments, added a small object detection head to enhance the capability for detecting small objects, incorporated an attention mechanism to increase the precision of object detection, and achieved ball detection in complex backgrounds on the sports field, which played a guiding role in the switching of broadcast camera angles.