Putri, Tsabita Rosyidah (2026) DETECTING THE RIPENESS OF PALM OIL FRUIT TO SUPPORT HARVESTING DECISIONS USING YOLO. Undergraduate thesis, UPN Veteran Jawa Timur.
|
Text (Cover)
22083010012.-cover.pdf Download (1MB) | Preview |
|
|
Text (Bab 1)
22083010012.-bab1.pdf Download (258kB) | Preview |
|
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (Bab 2)
22083010012.-bab2.pdf Restricted to Repository staff only until 19 May 2028. Download (1MB) |
|
|
Text (Bab 3)
22083010012.-bab3.pdf Restricted to Repository staff only until 19 May 2028. Download (669kB) |
|
|
Text (Bab 4)
22083010012.-bab4.pdf Restricted to Repository staff only until 19 May 2028. Download (1MB) |
|
|
Text (Bab 5)
22083010012.-bab5.pdf Download (239kB) | Preview |
|
|
Text (Daftar Pustaka)
22083010012.-daftarpustaka.pdf Download (193kB) | Preview |
|
|
Text (Lampiran)
22083010012.-lampiran.pdf Restricted to Repository staff only Download (251kB) |
Abstract
Oil palm is one of the leading commodities that plays an important role in the national economy, where the quality of the produced oil is strongly influenced by the ripeness level of Fresh Fruit Bunches (FFB). However, the determination of fruit ripeness in the field is still carried out manually through visual inspection, which is subjective, inconsistent, and prone to errors in the harvesting process. Therefore, an automated system is required to accurately and efficiently detect fruit ripeness levels. This study aims to develop an oil palm fruit ripeness detection system based on deep learning using an object detection approach, as well as to compare the performance of two models, namely YOLOv8 and YOLOv12. The dataset consists of oil palm fruit images annotated into three classes: ripe, semi-ripe, and unripe. The research process includes model training with identical hyperparameter configurations, evaluation using precision, recall, mean Average Precision (mAP50 and mAP50-95), confusion matrix analysis, and inference time measurement, as well as system implementation using a Streamlit-based web application. The results show that YOLOv8 outperforms YOLOv12 in overall performance. YOLOv8 achieved a precision of 0.974, recall of 0.987, mAP50 of 0.994, mAP50-95 of 0.892, with an average inference time of 31.28 ms. Meanwhile, YOLOv12 achieved a precision of 0.976, recall of 0.969, mAP50 of 0.990, mAP50-95 of 0.872, with an inference time of 35.69 ms. Overall, YOLOv8 demonstrates the best balance between detection accuracy, generalization ability, and computational efficiency. This study provides a contribution in the form of a recommended best-performing detection model and a web-based system implementation that can su support faster, more accurate, and more efficient decision-making in oil palm harvesting processes.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | Q Science > Q Science (General) Q Science > QA Mathematics Q Science > QA Mathematics > QA75 Electronic computers. Computer science S Agriculture > S Agriculture (General) T Technology > T Technology (General) T Technology > TP Chemical technology |
||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Data Science | ||||||||||||
| Depositing User: | Tsabita Rosyidah Putri | ||||||||||||
| Date Deposited: | 19 May 2026 06:23 | ||||||||||||
| Last Modified: | 19 May 2026 06:23 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/51902 |
Actions (login required)
 |
View Item |