PENGUJIAN ALGORITMA YOLO UNTUK DETEKSI CACAT PADA PRODUK HASIL ALUMUNIUM CASTING PADA INDUSTRI OTOMOTIF

This research addresses the challenges faced by one of the automotive parts manufacturing plants in Indonesia, in detecting leaks in car components manufactured through aluminum die casting. Existing manual monitoring methods are time-consuming, prone to human error, and pose risks to product quality and operational safety. To overcome these challenges, this research proposes the application of machine learning, specifically computer vision techniques, such as Object Detection to identify and localize gas bubbles using a Leak Tester Machine equipped with a camera sensor equipped with the YOLO algorithm to improve the efficiency of the detection process. As technology develops, there is a shift from manual approaches to more efficient automated systems, utilizing computer vision algorithms, hardware such as Nvidia Jetson and Digital Camera. The YOLO algorithm shows good accuracy in overcoming defect detection in aluminum casting products, achieving a precision of 0.94, recall of 0.82, and [email protected] of 0.90. The performance of the YOLO algorithm in detecting gas bubbles is 100 - 120 ms per frame from thestreaming camera so that the fps performance is 10 fps (frames per second).