Skripsi
SISTEM PENDETEKSI KUALITAS FISIK PADA BIJI KOPI SECARA REALTIME MENGGUNAKAN METODE YOU ONLY LOOK ONCE (YOLOV5)
In this global era, the demand for coffee beans has been increasing year by year. The economic value of coffee beans is greatly influenced by the quality of the beans being marketed. Manual assessment of coffee quality by experts or farmers tends to be subjective and time-consuming. Based on these conditions, this study developed software to detect coffee bean quality in real-time using the YOLOv5 method running on a smartphone device. This software aims to help consumers select coffee beans on the market by displaying labels indicating good and bad coffee bean quality. This research was conducted using three combinations of epoch and batch size, namely batch size 16 epoch 200, batch size 24 epoch 200, and batch size 32 epoch 200. Non-real-time testing was carried out by entering test data images, while real-time testing was conducted using a smartphone camera. In the real-time testing process, the best accuracy was obtained in the combination of batch size 32 and epoch 200 with an accuracy value of 92%. The non-real-time testing process obtained the best accuracy of 95%. Keywords: Coffee Beans, YOLOv5, Real-Time