IMPLEMENTASI ALGORITMA YOLOv11 PADA SISTEM KAMERA CERDAS BERBASIS WEBCAM UNTUK DETEKSI LUKA DIABETES

RAHMAT ILLAHI, . (2026) IMPLEMENTASI ALGORITMA YOLOv11 PADA SISTEM KAMERA CERDAS BERBASIS WEBCAM UNTUK DETEKSI LUKA DIABETES. Sarjana thesis, UNIVERSITAS NEGERI JAKARTA.

	Text Cover.pdf Download (3MB)
	Text BAB I.pdf Download (2MB)
	Text BAB II.pdf Restricted to Registered users only Download (3MB) | Request a copy
	Text BAB III.pdf Restricted to Registered users only Download (1MB) | Request a copy
	Text BAB IV.pdf Restricted to Registered users only Download (1MB) | Request a copy
	Text BAB V.pdf Restricted to Registered users only Download (576kB) | Request a copy
	Text DAFTAR PUSTAKA.pdf Download (1MB)
	Text Lampiran dan Daftar Riwayat Hidup.pdf Restricted to Registered users only Download (1MB) | Request a copy

Abstract

Luka diabetes merupakan salah satu komplikasi diabetes melitus yang dapat menyebabkan infeksi serius hingga amputasi apabila tidak ditangani sejak dini. Proses pemeriksaan luka yang masih dilakukan secara visual memiliki keterbatasan karena bergantung pada pengalaman tenaga kesehatan. Penelitian ini bertujuan untuk mengembangkan aplikasi sistem kamera cerdas berbasis webcam menggunakan algoritma YOLO11 untuk mendeteksi luka diabetes, menganalisis kinerja model berdasarkan metrik evaluasi, serta memvalidasi hasil deteksi menggunakan foto luka asli yang disertai data kadar gula darah sebagai data pendukung.Penelitian ini menggunakan metode Reuse-Oriented Software Engineering. Dataset citra luka diproses menggunakan Roboflow, kemudian model YOLO11 dilatih pada Google Colab dan diintegrasikan ke dalam aplikasi desktop DiabetScan yang dikembangkan menggunakan Python dan OpenCV. Sistem dilengkapi dengan fitur akuisisi citra menggunakan webcam, deteksi luka, penyimpanan hasil pemeriksaan ke Google Sheets, serta pengiriman hasil melalui WhatsApp.Hasil pelatihan model menunjukkan nilai precision sebesar 86,7%, recall sebesar 80,4%, F1-score sebesar 83,5%, mAP50 sebesar 84,2%, dan mAP50-95 sebesar 45,6%. Berdasarkan pengujian terhadap 15 data uji, sistem memperoleh accuracy sebesar 80%, dengan precision sebesar 87,5%, recall sebesar 77,8%, dan F1-score sebesar 82,4% untuk kelas Diabetic Ulcer. Selain itu, hasil validasi menggunakan foto luka asli yang disertai data kadar gula darah menunjukkan bahwa hasil prediksi sistem memiliki kesesuaian dengan label ground truth. Berdasarkan hasil tersebut, sistem yang dikembangkan mampu mendeteksi luka diabetes dengan baik dan berpotensi digunakan sebagai alat bantu skrining awal berbasis citra digital. ***** Diabetic wounds are one of the complications of diabetes mellitus that may lead to severe infections and lower-limb amputation if not treated at an early stage. Conventional visual examination of diabetic wounds has limitations because it depends on the experience of healthcare professionals. This study aims to develop a webcam-based smart camera application using the YOLO11 algorithm to detect diabetic wounds, analyze the model performance using evaluation metrics, and validate the detection results using real wound images accompanied by blood glucose measurements as supporting data.This study employed the Reuse-Oriented Software Engineering method. The wound image dataset was processed using Roboflow, while the YOLO11 model was trained on Google Colab and integrated into a desktop application called DiabetScan, developed using Python and OpenCV. The application provides image acquisition through a webcam, wound detection, automatic storage of examination results in Google Sheets, and result delivery via WhatsApp.The training results showed that the YOLO11 model achieved a precision of 86.7%, recall of 80.4%, F1-score of 83.5%, mAP50 of 84.2%, and mAP50-95 of 45.6%. Based on testing using 15 image samples, the system achieved an accuracy of 80%, with a precision of 87.5%, recall of 77.8%, and F1-score of 82.4% for the Diabetic Ulcer class. Furthermore, validation using real wound images accompanied by blood glucose examination results demonstrated that the system predictions were consistent with the corresponding ground truth labels. These results indicate that the proposed system is capable of detecting diabetic wounds effectively and has the potential to be used as an early screening tool based on digital wound images.

Item Type:	Thesis (Sarjana)
Additional Information:	1). Syufrijal, S.T., M.T. ; 2). Drs. Rimulyo Wicaksono, M.M.
Subjects:	Teknologi dan Ilmu Terapan > Teknik Elektronika
Divisions:	FT > D IV Teknologi Rekayasa Otomasi
Depositing User:	Users 35152 not found.
Date Deposited:	06 Aug 2026 03:22
Last Modified:	06 Aug 2026 03:22
URI:	http://repository.unj.ac.id/id/eprint/68698

Actions (login required)

View Item