| dc.description.abstract | Competition in the industrial world is getting tougher, the need to increase efficiency and accuracy in the quality inspection process is very important. CV Adi Makmur Metalindo is a palm oil machine component fabrication workshop that still applies manual quality inspection. Manual inspections are prone to errors, depend on human skills, and take a long time. This research aims to design a prototype machine learning-based quality inspection system to automatically detect defective products. The methodology used involves defective product image data collection, data labelling, and training using YOLO (You Only Look Once) based Convolutional Neural Network (CNN) algorithm. The prototype was implemented with an esp32-cam camera in performing defective product detection. The use of machine learning is able to identify defective products such as geometry defects, porous defects, and surface defects. Evaluation of model performance uses confusion matrix, loss graph, and precision-recall curve. The evaluation results show that the system can identify product defects with an accuracy of mAP50-95 of 74.5%, mAP50 of 88.5%, and the time required to detect 0.0084 seconds per image. The research proves that the use of machine learning in quality inspection can improve efficiency and reduce the dependence of manual inspection, thus strengthening the competitiveness of the company. | en_US |
