Oil palm loose fruit detection using YOLOv4 for an autonomous mobile robot collector

This study researches the usage of YOLOv4 for real-time loose fruit detection in oil palm plantations as the first step in implementing automation in the collection of loose fruits. Our system leverages high-resolution video data (4K and 1080p) from various plantation settings. To address the challe...

Full description

Bibliographic Details
Main Authors: Japar, Ahmed Fareed, Ramli, Hafiz Rashidi, Norsahperi, Nor Mohd Haziq, Wan Hasan, Wan Zuha
Format: Article
Language:English
Published: Institute of Electrical and Electronics Engineers Inc. 2024
Online Access:http://psasir.upm.edu.my/id/eprint/113876/
http://psasir.upm.edu.my/id/eprint/113876/1/113876.pdf
_version_ 1848866346769580032
author Japar, Ahmed Fareed
Ramli, Hafiz Rashidi
Norsahperi, Nor Mohd Haziq
Wan Hasan, Wan Zuha
author_facet Japar, Ahmed Fareed
Ramli, Hafiz Rashidi
Norsahperi, Nor Mohd Haziq
Wan Hasan, Wan Zuha
author_sort Japar, Ahmed Fareed
building UPM Institutional Repository
collection Online Access
description This study researches the usage of YOLOv4 for real-time loose fruit detection in oil palm plantations as the first step in implementing automation in the collection of loose fruits. Our system leverages high-resolution video data (4K and 1080p) from various plantation settings. To address the challenges of small and numerous loose fruits, we introduced an image preprocessing technique called “image tiling” into the vision system workflow and studied the effects this has on the performance of the detection model. This involves dividing the image into smaller sections for individual processing by both YOLOv4 and YOLOv4-tiny models, enhancing detection accuracy. Refined models (YOLOv4-tiling and YOLOv4-tiny-tiling) are then evaluated. YOLOv4 achieved the highest precision (97%) and F1-score (86.3%), while YOLOv4-tiling offered a slight improvement in recall (80.8%). Notably, YOLOv4-tiny, initially underperforming (precision: 37.2%, recall: 20.9%, F1-score: 25%), showed significant improvement with tiling (precision: 90.5%, recall: 67.1%, F1-score: 73.8%). Also, replacing the SPP layer in YOLOv4 with SPP-Fast resulted in increased precision (92.6%) and a significantly improved F1-score of 91.4%. This vision system was then integrated with a custom designed Loose Fruit Collector Robot through the Robot Operating System (ROS).
first_indexed 2025-11-15T14:19:09Z
format Article
id upm-113876
institution Universiti Putra Malaysia
institution_category Local University
language English
last_indexed 2025-11-15T14:19:09Z
publishDate 2024
publisher Institute of Electrical and Electronics Engineers Inc.
recordtype eprints
repository_type Digital Repository
spelling upm-1138762025-02-05T03:35:13Z http://psasir.upm.edu.my/id/eprint/113876/ Oil palm loose fruit detection using YOLOv4 for an autonomous mobile robot collector Japar, Ahmed Fareed Ramli, Hafiz Rashidi Norsahperi, Nor Mohd Haziq Wan Hasan, Wan Zuha This study researches the usage of YOLOv4 for real-time loose fruit detection in oil palm plantations as the first step in implementing automation in the collection of loose fruits. Our system leverages high-resolution video data (4K and 1080p) from various plantation settings. To address the challenges of small and numerous loose fruits, we introduced an image preprocessing technique called “image tiling” into the vision system workflow and studied the effects this has on the performance of the detection model. This involves dividing the image into smaller sections for individual processing by both YOLOv4 and YOLOv4-tiny models, enhancing detection accuracy. Refined models (YOLOv4-tiling and YOLOv4-tiny-tiling) are then evaluated. YOLOv4 achieved the highest precision (97%) and F1-score (86.3%), while YOLOv4-tiling offered a slight improvement in recall (80.8%). Notably, YOLOv4-tiny, initially underperforming (precision: 37.2%, recall: 20.9%, F1-score: 25%), showed significant improvement with tiling (precision: 90.5%, recall: 67.1%, F1-score: 73.8%). Also, replacing the SPP layer in YOLOv4 with SPP-Fast resulted in increased precision (92.6%) and a significantly improved F1-score of 91.4%. This vision system was then integrated with a custom designed Loose Fruit Collector Robot through the Robot Operating System (ROS). Institute of Electrical and Electronics Engineers Inc. 2024-08-21 Article PeerReviewed text en cc_by_nc_nd_4 http://psasir.upm.edu.my/id/eprint/113876/1/113876.pdf Japar, Ahmed Fareed and Ramli, Hafiz Rashidi and Norsahperi, Nor Mohd Haziq and Wan Hasan, Wan Zuha (2024) Oil palm loose fruit detection using YOLOv4 for an autonomous mobile robot collector. IEEE Access, 12. pp. 138582-138593. ISSN 2169-3536; eISSN: 2169-3536 https://ieeexplore.ieee.org/document/10643084/ 10.1109/ACCESS.2024.3446890
spellingShingle Japar, Ahmed Fareed
Ramli, Hafiz Rashidi
Norsahperi, Nor Mohd Haziq
Wan Hasan, Wan Zuha
Oil palm loose fruit detection using YOLOv4 for an autonomous mobile robot collector
title Oil palm loose fruit detection using YOLOv4 for an autonomous mobile robot collector
title_full Oil palm loose fruit detection using YOLOv4 for an autonomous mobile robot collector
title_fullStr Oil palm loose fruit detection using YOLOv4 for an autonomous mobile robot collector
title_full_unstemmed Oil palm loose fruit detection using YOLOv4 for an autonomous mobile robot collector
title_short Oil palm loose fruit detection using YOLOv4 for an autonomous mobile robot collector
title_sort oil palm loose fruit detection using yolov4 for an autonomous mobile robot collector
url http://psasir.upm.edu.my/id/eprint/113876/
http://psasir.upm.edu.my/id/eprint/113876/
http://psasir.upm.edu.my/id/eprint/113876/
http://psasir.upm.edu.my/id/eprint/113876/1/113876.pdf