Summary
International Workshop on Smart Info-Media Systems in Asia
2022
Session Number:SS2
Session:
Number:SS2-2
Study of Oil Palm Stalk Detection based on YOLOv4
Jin Wern Lai, Hafiz Rashidi Ramli, Luthffi Idzhar Ismail, Wan Zuha Wan Hasan,
pp.69-72
Publication Date:2022/9/15
Online ISSN:2188-5079
PDF download
Summary:
Fresh fruit bunch (FFB) is the main component to produce palm oil. Harvesting FFB efficiently within the optimal period is crucial to achieving the maximum oil extraction rate (OER) and quality. In this study, as a novelty, a stalk detection system is developed as a precursor to an autonomous FFB harvesting system. A depth camera is used and the system operates with a Convolutional Neural Network (CNN) known as YOLOv4. To train the YOLOv4 model, the data is collected by recording videos using Realsense Camera D435 with 1080p resolution. Intel Core i7-8750H processor and GeForce DTX 1070 graphics card are used during the training process. The result of mean average precision (mAP) is 29.67% and 20.05% for the dataset without LED light and with light-emitting diode (LED).