PERANCANGAN SISTEM DETEKSI DAN PEMANTAUAN KONDISI TELUR PADA ALAT PENETAS TELUR MENGGUNAKAN ALGORITMA YOLO

Nabila Nur Banafsyah, - and Diky Zakaria, - and Muhammad Rizalul Wahid, - (2025) PERANCANGAN SISTEM DETEKSI DAN PEMANTAUAN KONDISI TELUR PADA ALAT PENETAS TELUR MENGGUNAKAN ALGORITMA YOLO. S1 thesis, Universitas Pendidikan Indonesia.

Abstract

Kebutuhan akan efisiensi penetasan telur unggas meningkat seiring dengan pertumbuhan industri peternakan. Alat penetas telur konvensional memiliki keterbatasan dalam pemantauan kondisi telur secara real-time, yang dapat menyebabkan kerugian akibat kegagalan penetasan atau kematian anakan ayam. Permasalahan ini sering kali diperburuk oleh ketidakmampuan sistem pendeteksi gerak konvensional untuk mengidentifikasi anakan ayam yang baru menetas secara akurat, sehingga penanganan pasca-penetasan menjadi terlambat. Penelitian ini bertujuan mengembangkan sistem deteksi dan pemantauan kondisi telur ayam pada alat penetas telur menggunakan algoritma object detection YOLOv5 dan YOLOv8 untuk meningkatkan efisiensi penetasan. Metode penelitian menggunakan pendekatan R&D dengan model ADDIE, meliputi tahap analisis, perancangan, pengembangan, implementasi, dan evaluasi. Dataset berjumlah 2.122 citra telur (retak dan tidak retak) diperoleh dari pengambilan gambar langsung dan sumber publik, dengan variasi pencahayaan dan posisi. Lima varian model YOLOv5 dan YOLOv8 dilatih dan diuji menggunakan metrik precision, recall, dan mAP@0.5. Hasil pengujian menunjukkan YOLOv8n unggul pada dataset (precision 99,3%, recall 98,6%, mAP@0.5 99,4%), namun pada implementasi real-time di inkubator mengalami false positive hingga ±12%. YOLOv5l dipilih sebagai model terbaik karena stabil di perangkat berspesifikasi menengah dengan false positive <5%. Model ini diintegrasikan dengan sistem kendali suhu, kelembapan, dan egg turning, menghasilkan keberhasilan 100% pada uji fungsionalitas. Sistem ini efektif memantau kondisi telur secara real-time dan mendukung peningkatan keberhasilan penetasan. ----- The need for efficient poultry egg hatching is increasing in line with the growth of the livestock industry. Conventional egg incubators have limitations in monitoring egg conditions in real time, which can lead to losses due to hatching failure or chick mortality. This issue is often exacerbated by the inability of conventional motion detection systems to accurately identify newly hatched chicks, resulting in delayed post-hatching handling. This study aims to develop an egg condition detection and monitoring system for an incubator using YOLOv5 and YOLOv8 object detection algorithms to improve hatching efficiency. The research employed an R&D approach with the ADDIE model, comprising analysis, design, development, implementation, and evaluation stages. A dataset of 2,122 images (cracked and uncracked eggs) was collected from direct camera capture and public sources, with variations in lighting and position. Five variants of YOLOv5 and YOLOv8 were trained and evaluated using precision, recall, and mAP@0.5 metrics. Results show that YOLOv8n achieved the highest dataset performance (precision 99.3%, recall 98.6%, mAP@0.5 99.4%), but in real-time incubator testing, it produced up to ±12% false positives. YOLOv5l was selected as the best model due to its stability on mid-spec devices, with false positives under 5%. The model was integrated with temperature, humidity, and egg-turning controls, achieving 100% functionality in testing. The system effectively monitors egg conditions in real-time and supports improved hatching success.

	Text S_MKB_2109471_Title.pdf Download (462kB)
	Text S_MKB_2109471_Chapter1.pdf Download (297kB)
	Text S_MKB_2109471_Chapter2.pdf Restricted to Staf Perpustakaan Download (1MB)
	Text S_MKB_2109471_Chapter3.pdf Download (518kB)
	Text S_MKB_2109471_Chapter4.pdf Restricted to Staf Perpustakaan Download (1MB)
	Text S_MKB_2109471_Chapter5.pdf Download (318kB)
	Text S_MKB_2109471_Appendix.pdf Restricted to Staf Perpustakaan Download (634kB)

Item Type:	Thesis (S1)
Additional Information:	https://scholar.google.com/citations?view_op=new_articles&hl=en&imq=Nabila+Nur+Banafsyah# ID SINTA Dosen Pembimbing: Diky Zakaria : 6779007 Muhammad Rizalul Wahid : 6780434
Uncontrolled Keywords:	Alat Penetas Telur, Object Detection, YOLOv5, YOLOv8, Computer Vision Egg Incubator, Object Detection, YOLOv5, YOLOv8, Computer Vision
Subjects:	T Technology > T Technology (General)
Divisions:	UPI Kampus Purwakarta > S1 Mekatronika dan Kecerdasan Buatan
Depositing User:	Nabila Nur Banafsyah
Date Deposited:	18 Sep 2025 06:17
Last Modified:	18 Sep 2025 06:17
URI:	http://repository.upi.edu/id/eprint/139846

Actions (login required)

View Item