Pelacakan Bola Dan Pemain Pada Pertandingan Sepak Bola Menggunakan Model YOLOv11

Dewantara, Kevin Marco (2025) Pelacakan Bola Dan Pemain Pada Pertandingan Sepak Bola Menggunakan Model YOLOv11. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5002211134-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (15MB) | Request a copy

Abstract

Sepak bola merupakan olahraga yang sangat populer di dunia dan memiliki potensi besar dalam pengembangan analitik berbasis visi komputer. Penelitian ini bertujuan untuk mengembangkan sistem pelacakan bola dan pemain pada pertandingan sepak bola menggunakan model You Only Look Once version 11 (YOLOv11). Dataset yang digunakan terdiri dari citra pertandingan sepak bola yang telah dianotasi ke dalam lima kelas objek, yaitu player, goalkeeper, referee, ball, dan background. Model dilatih selama 90 epoch dan menghasilkan nilai evaluasi berupa mAP@0.5 sebesar 0.727 dan mAP@0.5:0.95 sebesar 0.48. Setelah pelatihan, model diuji pada dua skenario video pertandingan dengan karakteristik berbeda, untuk mengevaluasi kemampuan pelacakan dan estimasi interaksi antara bola dan pemain. Hasil pengujian menunjukkan bahwa model mampu mendeteksi pemain dan bola dengan akurasi yang tinggi, serta menunjukkan potensi dalam analisis taktis seperti estimasi penguasaan bola. Penelitian ini menunjukkan bahwa pendekatan deteksi objek berbasis YOLOv11 dapat digunakan secara efektif dalam sistem pelacakan otomatis untuk mendukung analisis pertandingan sepak bola.
======================================================================================================================================
Football is a highly popular sport worldwide, with significant potential for the development of computer vision-based analytics. This study aims to develop a ball and player tracking system in football matches using the You Only Look Once version 11 (YOLOv11) model. The dataset used consists of football match images that have been annotated into five object classes, namely player, goalkeeper, referee, ball, and background. The model was trained for 90 epochs and produced evaluation values in the form of mAP@0.5 of 0.727 and mAP@0.5:0.95 of 0.48. After training, the model was tested on two match video scenarios with different characteristics to evaluate its tracking capabilities and interaction estimation between the ball and players. The test results show that the model can detect players and the ball with high accuracy, and shows potential in tactical analysis such as ball possession estimation. This study demonstrates that the YOLOv11-based object detection approach can be effectively used in an automated tracking system to support soccer match analysis.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	YOLOv11, pelacakan objek, sepak bola, visi komputer, object detection, football, computer vision, football
Subjects:	Q Science > QA Mathematics > QA336 Artificial Intelligence
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User:	Kevin Marco Dewantara
Date Deposited:	01 Aug 2025 06:28
Last Modified:	01 Aug 2025 06:28
URI:	http://repository.its.ac.id/id/eprint/125871

Actions (login required)

View Item