Real-Time Autonomous Detection and Localization of Loose Fruits in Oil Palm Plantations Using YOLOv4 and RGB-D

DOI

10.2991/978-94-6463-566-9_5How to use a DOI?

Keywords

Loose oil palm fruit; autonomous detection; YOLO; RGB-D

Abstract

Loose Palm Fruit (LPF) is an oil palm fruit that has ripened and fallen from its bunch, containing high oil content. Each loss of LPF affects the oil extraction rate and results in financial losses. Existing LPF collection methods are not very effective as they require human control and supervision. Conventional methods, such as mechanical and roller-type LPF collectors, are inefficient because LPF is scattered over extensive plantations. Therefore, an autonomous LPF detection system is necessary. However, image-based detection systems are often disturbed by environmental factors such as brightness and grass, and the LPF location changes with the robot and camera position. The general objective of this study is to develop an accurate and efficient image-based LPF detection algorithm. This requires an efficient detection algorithm for real-time applications based on deep learning. Additionally, accurately determining the LPF location using image depth (RGB-D) is essential. This project employs a YOLOv4 object detector with high efficiency and accuracy to achieve real-time LPF detection. The LPF location is determined through the distance between the center coordinates of the LPF bounding box and the camera using depth images and the horizontal field of view of the Intel RealSense D435i camera. This system is integrated into the Robot Operating System (ROS) to ensure usability in robots. The system achieved a Mean Accuracy (mAP@IoU 0.5) of 98.74%, an average loss of 0.124, and a detection time of 5.14ms. For LPF location determination, the difference between the algorithm’s calculated locations and manual measurements is only 3.82cm for the X coordinate and 1.80cm for the Y coordinate.

Copyright

© 2024 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Lee Teng Ching
AU  - Aqilah Baseri Huddin
AU  - Fazida Hanim Hashim
AU  - Mohd Faisal Ibrahim
PY  - 2024
DA  - 2024/11/01
TI  - Real-Time Autonomous Detection and Localization of Loose Fruits in Oil Palm Plantations Using YOLOv4 and RGB-D
BT  - Proceedings of the  International Conference on Advanced Technology and Multidiscipline (ICATAM 2024)
PB  - Atlantis Press
SP  - 54
EP  - 65
SN  - 2352-5401
UR  - https://doi.org/10.2991/978-94-6463-566-9_5
DO  - 10.2991/978-94-6463-566-9_5
ID  - Ching2024
ER  -