CD Skripsi

Deteksi Cacat Pada Tandan Buah Segar Kelapa Sawit Menggunakan Computer Vision Berbasis Algoritma You Only Look Once

---

ABSTRACT
Every year oil palm fresh fruit bunches (FFB) production continues to increase. However, one of the problems in crude palm oil (CPO) production is the process of sorting and grading, which is still done manually and traditionally. This problem can be eased by introducing a computer vision-based sorting and grading system and you only look once (YOLO) object detection method. This study aimed to build a YOLO object detection system for defective FFB based on the RGB values of each object and compared with normal oil palm FFB. The computer vision system consists of an RGB camera, a conveyor, and a python-based computer program with YOLO object detection. This system analyzed the external quality of FFB based on Minister of Agriculture Regulation No. 14 of 2013, including rotten oil palm, thorny FFB, empty, and long stalk FFB. The YOLO detection system was tested in four-room lighting conditions with lux values of 215, 305, 306, and 317. The test result of the detection model with the highest accuracy was obtained at a lux of 317 with an accuracy of oil palm FFB normal 99%, 82% for thorny FFB, 98% for rotten, 95% for empty FFB, and 98% for long stalk FFB. The accuracy validation of the object detection system used the confusion matrix which resulted in mAP of 95%. The YOLO object detection method using video frames can make it easier to sort moving FFB on the conveyor and obtain RGB values using a python program to determine how well the object detection system can recognize objects.
Keywords: Fresh fruit bunches (FFB), sorting and grading, defect, computer vision, YOLO, confusion matrix.

---

Ketersediaan

Informasi Detail

Judul Seri

-

No. Panggil

1803112139

Penerbit

Pekanbaru : Universitas Riau - FMIPA - Fisika., 2023

Deskripsi Fisik

-

Bahasa

Indonesia

ISBN/ISSN

-

Klasifikasi

1803112139

Tipe Isi

-

Tipe Media

-

Tipe Pembawa

-

Edisi

-

Subjek

FISIKA

Info Detail Spesifik

-

Pernyataan Tanggungjawab

tethi

Versi lain/terkait

Lampiran Berkas

Komentar

---

Anda harus masuk sebelum memberikan komentar