ANALISIS PERBANDINGAN KINERJA YOLO DAN CAMSHIFT DALAM PELACAKAN OBJEK BERBASIS VIDEO

Abstract

This study aims to implement and analyze the YOLO (You Only Look Once) method for human object tracking and compare it with the Camshift (Continuously Adaptive Mean Shift) method. Supported by the rapid development of artificial intelligence (AI) in information technology, this research explores the capabilities of both methods in object tracking. The process involves video capture, tracking using YOLO and Camshift, and performance evaluation through the Intersection over Union (IoU) metric under various lighting conditions. Both methods are integrated into an Arduino-based tracking device. The results show that YOLO outperforms in environments with complex backgrounds and optimal lighting, although it is slower, while Camshift is faster but less accurate under varying lighting conditions. Both methods are effective in monitoring human movement, but there is a trade-off between accuracy and speed. With superior efficiency and accuracy, YOLO is more suitable for real-time human object tracking. Future research is suggested to combine or enhance detection and tracking algorithms, optimize the system with advanced hardware, test under real-world conditions, and explore the integration of other technologies to create a more reliable and adaptive tracking system.