Home // SIGNAL 2026, The Eleventh International Conference on Advances in Signal, Image and Video Processing // View article

YOLO-Based Deep Learning Models for Real-Time Volleyball Action Recognition

Authors:
Brandon Labio
Atul Dubey

Keywords: Volleyball action detection; deep learning; YOLO11; YOLOv8

Abstract:
Analyzing player performance through highlights and statistics is critical for volleyball coaching, but remains a labor-intensive bottleneck. Current methods are often hindered by impractical deployment, inefficiency, and lack of portability. To overcome these limitations, this study introduces a transportable, deep-learning-based object detection model for realtime identification of volleyball actions. Using a dataset of 1,236 images, models were trained based on both YOLOv8 and YOLO11. The YOLOv8 model yielded superior performance, achieving a mAP50 value of 87.8%, and was optimized at a learning rate of 0.0005 over 80 epochs. This model was subsequently tested using the test dataset and achieved a mAP50 of 79.8%. The system was then deployed on a Raspberry Pi to evaluate real-time feasibility, reaching a latency of 1 FPS. These findings establish a critical baseline for the development of live automated highlight generation systems and are essential for efficient creation of highlights and analysis of performance.

Pages: 8 to 13

Copyright: Copyright (c) IARIA, 2026

Publication date: March 8, 2026

Published in: conference

ISSN: 2519-8432

ISBN: 978-1-68558-360-6

Location: Valencia, Spain

Dates: from March 8, 2026 to March 12, 2026