SMART FOOTBALL: APPLICATION OF OBJECT DETECTION TO IDENTIFY OBJECTS IN SOCCER MATCH VIDEOS

• Main Author: Alif Syahreza, Zachrandika
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/76659
• Institution: Institut Teknologi Bandung

Soccer is a highly popular sport worldwide. Indonesia is one of the countries with a high enthusiasm for soccer among its people. However, the development of Indonesia’s soccer achievements is less than satisfactory due to several problem factors. One of the issues leading to the lack of success in Indonesian soccer is the inadequate supporting facilities for player training. Thus, in this Final Project, a YOLOv8 object detection model will be built using transfer learning with a pretrained YOLOv8x model. Several experiments will be conducted implementing the model using different optimizer aspects, namely SGD, AdamW, and Adam. Then a comparison will be made among object detection models based on the highest mAP value and the fastest inference time. The models based on the highest mAP value is YOLOv8x – SGD, and the model with the fastest inference time is YOLOv8x – Adam. Next, a comparison will be made between these two models by implementing object tracking and classifying player objects based on jersey color using HSV color filtering method on a 38 second video of an Indonesian League soccer match in 68 seconds.