CD Skripsi
Rancang Bangun Prototype Pemilah Brondolan Kelapa Sawit Otomatis Berbasis Kamera Dan Arduino Uno
The development of digital technology has driven a major transformation in the manufacturing industry. This study designed and built a prototype of an automatic loosed fruits sorter based on a camera and Arduino Uno. The system consists of two main components that are integrated with each other, namely the detection system and the automation system. The detection system includes the preparation of tools and materials, data acquisition and annotation to build a dataset, the development of a YOLO detection model, and model training and testing. The dataset consists of three classes of palm kernel: fresh, rotten, and dry. The model uses the YOLOv8-s algorithm, which was trained and tested to detect these three classes using Google Colab. Model performance evaluation was conducted using a confusion matrix to measure accuracy, precision, recall, and F1-Score. The results showed that for the fresh class, precision reached 97.5%, recall 100%, and F1-score 98.7%. For the rotten and dry classes, precision, recall, and F1-score each reached 100%. The overall accuracy of the detection system was 100% with a mean Average Precision (mAP) value of 99.5%. The automation system was designed using Arduino Uno and MG996R servo motors to move a plywood arm as an automatic separator. The YOLOv8-s model is integrated in real-time with the automation system. Test results show that the sorting accuracy for the fresh class is 88%, rotten 70%, and dry 74%. These results indicate that the prototype can be used to separate loose fruit based on these three class categories.
Keywords : Machine Vision, Oil Palm Fruit, YOLO Algorithm Arduino Uno, Classification.
Tidak tersedia versi lain