Project AURIONAI Use-Cases Research & Innovation
Agriculture

Fresh Fruit Bunch Ripeness Grading

Object detectors grading palm oil FFB, at mills or in apps.

Named Models

Gunadarma UniversityYOLOv8 FFB ripeness grader (Gunadarma)Gunadarma UniversityReleasedDec 2025Runs onnot in the notesClassvisionAcademic FFB grading model in a Streamlit web app, trained on 6,592 Roboflow images; precision 94.5%, recall 94.7%, mAP 98%. No plantation deployment stated.Commercial use: not stated; Jakarta hosting: unknown; Indonesian support: not statedUPN Veteran Jawa TimurYOLOv11 FFB ripeness grader (IJCCS 2025)UPN Veteran Jawa TimurReleasedOct 2025Runs onCPU server, offline capableClassvisionAcademic oil palm fresh fruit bunch ripeness detector on YOLOv11; 90.2% mAP@50. No plantation deployment stated. Affiliation unverified.Commercial use: not stated; Jakarta hosting: unknown; Indonesian support: not statedPalm oil FFB ripeness (zah, Roboflow)not in the notesReleasednot in the notesRuns onnot in the notesClassvisionCommunity oil palm fresh fruit bunch ripeness detection project on Roboflow Universe: about 7,070 images with Indonesian ripeness classes (mentah, kurang masak, masak, terlalu masak), 1 trained model.Commercial use: not stated; Jakarta hosting: unknown; Indonesian support: not statedPalm Oil Ripeness (braincore, Roboflow)not in the notesReleasednot in the notesRuns onnot in the notesClassvisionCommunity oil palm fruit ripeness detection project on Roboflow Universe: about 3,940 images, 3 trained models.Commercial use: not stated; Jakarta hosting: unknown; Indonesian support: not statedSawit ripeness (testing workspace, Roboflow)not in the notesReleasednot in the notesRuns onnot in the notesClassvisionCommunity oil palm bunch grading project on Roboflow Universe: about 3,940 images with classes including Janjang Kosong, Kurang Masak, Masak, Mentah, Terlalu Masak and Abnormal; 2 trained models.Commercial use: not stated; Jakarta hosting: unknown; Indonesian support: not stated
Where this comes from
Models the research names for this use case. Research, with sources on each model record.

Who Runs It

production, 1research, 1planned, 1
PKS PGK Jambi (palm oil mill; owner not stated)Country: IndonesiaAI-based FFB grading system integrated with mySAP365 (model not stated)Vendor: mySAP365 Agrobusiness ERP platform; developer not statedproductionDate24 Sep 2025Implemented at the mill per PalmCo release; audit trail from weighbridge to payment; PalmCo called it suitable for replicationEvidence: company statementConfidence: lowSource: palmoilmagazine.com
PT Toba Agro Manunggal with BINUS (BDSRC)Country: IndonesiaSmartSawitGrading, lightweight deep learning FFB ripeness grading on smartphones (architecture not named)Vendor: BINUS University (BDSRC)researchDate28 Jul 2026Developed; rollout not statedEvidence: company statementConfidence: lowSource: research.binus.ac.id
PTPN IV PalmCoPTPN IV PalmCoCountry: IndonesiaBenchmark of the PGK FFB grading systemVendor: not statedplannedDate24 Sep 2025Benchmark visit for possible rollout to PalmCo mills; no PalmCo deployment statedEvidence: company statementConfidence: lowSource: palmoilmagazine.com, koran-jakarta.com
Where this comes from
Documented uses by named organisations. The evidence column says who made the claim.

Candidate Models

UltralyticsYOLO26lUltralytics, visionReleasedJan 2026open weightsruns on edgehas scoreslast 18 monthsUltralyticsYOLO26mUltralytics, visionReleasedJan 2026open weightsruns on edgehas scoreslast 18 monthsUltralyticsYOLO26nUltralytics, visionReleasedJan 2026open weightsruns on edgehas scoreslast 18 monthsUltralyticsYOLO26sUltralytics, visionReleasedJan 2026open weightsruns on edgehas scoreslast 18 monthsUltralyticsYOLO26xUltralytics, visionReleasedJan 2026open weightshas scoreslast 18 monthsRoboflowRF-DETR 2XLRoboflow, visionReleasedNov 2025runs on edgehas scoreslast 18 monthsRoboflowRF-DETR MRoboflow, visionReleasedNov 2025runs on edgehas scoreslast 18 monthsRoboflowRF-DETR NRoboflow, visionReleasedNov 2025runs on edgehas scoreslast 18 monthsRoboflowRF-DETR SRoboflow, visionReleasedNov 2025runs on edgehas scoreslast 18 monthsMetaSAM 3Meta, visionReleasedNov 2025open weightshas scoreslast 18 months

Showing 10 of 655. All models of this class

Where this comes from
Derived, not researched: models in the directory whose class matches a class this use case uses, excluding family records and models shut down or deprecated. Ranked by how many fit facts they meet, then by release date. Each fit fact is read from the model record.

Sources

  1. Optimization of Palm Fruit Ripeness Detection With Yolov11 on CPU | Anniswa | IJCCS (Indonesian Journal of Computing and Cybernetics Systems journal.ugm.ac.id https://journal.ugm.ac.id/ijccs/article/view/111253
  2. Optimalisasi Deteksi Tingkat Kematangan Tanda Buah Segar Kelapa Sawit Menggunakan YOLOV8 Dengan Platform Web | Jurnal Ilmiah Teknologi dan R ejournal.gunadarma.ac.id https://ejournal.gunadarma.ac.id/index.php/tekno/article/view/67
  3. Palm Oil Magazine - PalmCo PTPN 4 Benchmarks AI-Powered FFB Grading System at PGK Jambi Mill - Palm Oil Magazine palmoilmagazine.com https://www.palmoilmagazine.com/palmco/2025/10/02/palmco-ptpn-4-benchmarks-ai-powered-ffb-grading-system-at-pgk-jambi-mill/

To cite this page: Project AURION, AI Use-Cases Research & Innovation, "Fresh Fruit Bunch Ripeness Grading", https://aurion.semai.id/usecases/ag-ffb-ripeness-grading/, records of October 2026.