Sistem Deteksi Dan Penandaan Fertilitas Telur Ayam Pada Proses Candling Menggunakan Metode YOLOv8n

Oktavia, Fita Dwi (2026) Sistem Deteksi Dan Penandaan Fertilitas Telur Ayam Pada Proses Candling Menggunakan Metode YOLOv8n. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2040221042-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (6MB) | Request a copy

Abstract

PT. XYZ menggunakan telur ayam fertil sebagai media biologis dalam proses produksi vaksin inaktif. Pemeriksaan fertilitas telur ayam dilakukan menggunakan mesin candling dengan pengolahan citra berdasarkan karakteristik warna atau intensitas piksel. Namun, variasi tampilan telur dan keterbatasan penetrasi cahaya menyebabkan akurasi deteksi masih berada pada kisaran 50–60%, sehingga sebagian telur perlu diperiksa ulang secara manual. Selain itu, penandaan telur infertil masih dilakukan oleh operator sehingga berpotensi menimbulkan kesalahan, seperti telur terlewat, tertukar, atau tercampur kembali dengan telur fertil. Penelitian ini menerapkan sistem deteksi fertilitas telur menggunakan YOLOv8n serta mekanisme penandaan otomatis pada telur infertil. Dataset yang digunakan terdiri atas 2.557 citra hasil candling dengan 20.209 anotasi bounding box yang terbagi ke dalam kelas fertil dan infertil. Keluaran model berupa kelas objek, nilai confidence, dan koordinat bounding box. Titik pusat bounding box digunakan untuk menentukan posisi telur pada candling slot C1 sampai C9. Apabila telur terdeteksi sebagai infertil, informasi posisi dikirimkan kepada Arduino Nano untuk menggerakkan motor stepper pada sumbu X, Y, dan Z menuju lokasi penandaan. Hasil evaluasi model menunjukkan nilai precision sebesar 0,9956, recall sebesar 1,00, F1-score sebesar 0,9978, dan mAP50 sebesar 0,9950. Pada pengujian deteksi menggunakan prototipe, sistem menghasilkan 265 deteksi benar dari 270 pengamatan dengan akurasi sebesar 97,41%. Pada pengujian penandaan, seluruh telur infertil yang menjadi target berhasil diberi tanda sesuai dengan posisinya. Hasil tersebut menunjukkan bahwa sistem mampu meningkatkan akurasi deteksi fertilitas telur ayam dan melakukan penandaan otomatis pada telur infertil sesuai dengan posisi slot candling.
=============================================================================================================================================
PT. XYZ uses fertilized chicken eggs as a biological medium in the production of inactivated vaccines. The fertility of Specific Antibody Negative (SAN) chicken eggs is examined using a candling machine with image processing based on color characteristics or pixel intensity. However, variations in egg appearance and limited light penetration result in a detection accuracy of only 50–60%, requiring some eggs to be manually re-examined. In addition, infertile eggs are still marked manually by operators, which may cause errors such as missed eggs, swapped eggs, or infertile eggs being mixed back in with fertile eggs. This study develops an egg fertility detection system using YOLOv8n integrated with an automatic marking mechanism for infertile eggs. The dataset consists of 2,557 candling images with 20,209 bounding-box annotations divided into fertile and infertile classes. The model outputs the object class, confidence score, and bounding-box coordinates. The center point of each bounding box is used to determine the egg position in candling slots C1 through C9. When an egg is detected as infertile, its position information is sent to an Arduino Nano, which controls stepper motors on the X, Y, and Z axes to move the marking mechanism to the target position. The model evaluation achieved a precision of 0.9956, recall of 1.00, F1-score of 0.9978, and mAP50 of 0.9950. During prototype detection testing, the system produced 265 correct detections out of 270 observations, resulting in an accuracy of 97.41%. During the marking test, all target infertile eggs were successfully marked at their correct positions. These results demonstrate that the developed system can improve the accuracy of chicken egg fertility detection and automatically mark infertile eggs according to their candling slot positions.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Candling, Convolutional Neural Network (CNN), egg fertility detection,automatic marking, YOLOv8n,Candling, Convolutional Neural Network (CNN), deteksi fertilitas telur, penandaan otomatis, YOLOv8n,
Subjects:	T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T59.7 Human-machine systems. T Technology > TJ Mechanical engineering and machinery > TJ212 Control engineering systems. Automatic machinery (General) T Technology > TJ Mechanical engineering and machinery > TJ223.A25 Actuators. T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7882.P3 Pattern recognition systems
Divisions:	Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User:	Fita Dwi Oktavia
Date Deposited:	05 Aug 2026 04:34
Last Modified:	05 Aug 2026 04:34
URI:	http://repository.its.ac.id/id/eprint/143994

Actions (login required)

View Item