Pemilahan Tanaman Kaktus oleh Robot Lengan Menggunakan Algoritma YOLOv3

Abstract

Science and technology continue to rapidly develop, including in the field of computers, which is divided into several branches such as computer vision and robotics. Computer vision studies, analyzes, and interprets visual information in the form of videos or images, and YOLO (You Only Look Once) is one of the most frequently used algorithms due to its fast and accurate detection capabilities. YOLOv3 is considered a stable and easy-to-use version of YOLO, but it produces larger and more complex files. Meanwhile, in robotics, robotic arms are commonly used in industries, but their high cost can be overcome by using 3D-printed robotic arms. On the other hand, edge computing can be used for better and cheaper computational processes, which have the potential to produce an automatic object identification and sorting system for cactus plants at an economic cost. The results of this research are obtained in the form of accuracy, recall, precision, F1 score, and mAP. The total accuracy value is 78%, recall is 96%, precision is 80%, F1 score is 97%, and mAP is 83% out of a total of 543 data. It can be concluded that cactus plant sorting using a combination of YOLOv3 and robotic arms has been successfully implemented and is quite effective in sorting objects.