Deteksi Objek pada Citra Menggunakan Model YOLO

Abstract

Object detection is a crucial task in the field of computer vision and digital image processing, with numerous practical applications. This paper focuses on the implementation of the You Only Look Once (YOLO) model, a deep learning-based approach for object detection. The YOLO model offers several advantages over previous methods, such as simultaneous prediction of bounding boxes and object class probabilities, a relatively simple Convolutional Neural Network (CNN) architecture, and high computational speed, making it suitable for real-time applications. The study utilizes a dataset of 770 images, with 524 for training, 136 for validation, and 110 for testing, specifically focused on detecting various pet animals. The training process involves annotation of the image data, followed by training and validation of the YOLO model. The results demonstrate the model's ability to effectively detect and classify objects, achieving high performance metrics such as precision, recall, and mean Average Precision (mAP) nearing 0.8 towards the end of the training process. Additionally, a confusion matrix is presented, highlighting the model's accuracy in classifying different classes, with the highest accuracy for the 'Cat' class at 95%. The paper concludes by discussing the model's performance and potential areas for improvement.