Implementasi Convolutional Neural Network Untuk Klasifikasi Penyakit Kulit Berbasis Citra Menggunakan Mobilenetv2

Kolbi, Zakia (2026) Implementasi Convolutional Neural Network Untuk Klasifikasi Penyakit Kulit Berbasis Citra Menggunakan Mobilenetv2. Other thesis, InstitutTeknologi Sepuluh Nopember.

Text 5025211049-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Perkembangan teknologi pengolahan citra dan kecerdasan buatan membuka peluang pemanfaatan citra digital sebagai media pendukung dalam klasifikasi penyakit kulit. Tugas akhir ini bertujuan untuk menerapkan metode Convolutional Neural Network (CNN) dengan arsitektur MobileNetV2 dalam klasifikasi penyakit kulit berbasis citra. Dataset yang digunakan terdiri dari delapan kelas, yaitu Acne, Carcinoma, Eczema, Keratosis, Milia, Normal, Other, dan Rosacea. Pada tahap awal, citra masukan diproses melalui tahapan preprocessing yang meliputi resize ke ukuran 224×224 piksel, normalisasi, serta penerapan data augmentation untuk meningkatkan variasi data latih dan mengurangi risiko overfitting. Model MobileNetV2 memanfaatkan bobot pralatih sebagai feature extractor dan dilakukan fine-tuning pada beberapa lapisan konvolusi terakhir. Evaluasi kinerja model dilakukan menggunakan metrik accuracy, precision, recall, dan F1-score. Hasil pengujian menunjukkan bahwa model MobileNetV2 mampu mencapai nilai accuracy sebesar 97,75%, dengan nilai precision 97,76%, recall 97,75%, dan F1-score 97,74%. Model yang telah dilatih kemudian diimplementasikan dalam arsitektur client–server menggunakan FastAPI sebagai backend layanan inferensi dan Flutter sebagai antarmuka aplikasi mobile. Hasil implementasi menunjukkan bahwa sistem mampu melakukan klasifikasi penyakit kulit berbasis citra secara near real-time melalui perangkat mobile.
========================================================================================================================
Advances in image processing and artificial intelligence have enabled the use of digital images as supporting media for skin disease classification. This study aims to implement a Convolutional Neural Network (CNN) using the MobileNetV2 architecture for image-based skin disease classification. The dataset used in this research consists of eight classes, namely Acne, Carcinoma, Eczema, Keratosis, Milia, Normal, Other, and Rosacea. In the initial stage, input images undergo preprocessing steps including resizing to 224×224 pixels, normalization, and data augmentation to increase training data diversity and reduce the risk of overfitting. The MobileNetV2 model utilizes pre-trained weights as a feature extractor, with finetuning applied to the last convolutional layers. Model performance is evaluated using accuracy, precision, recall, and F1-score metrics. The experimental results show that the MobileNetV2 model achieves an accuracy of 97.75%, with precision, recall, and F1-score values of 97.76%, 97.75%, and 97.74%, respectively. The trained model is then deployed using a client–server architecture, where FastAPI serves as the backend inference service and Flutter is used as the mobile application interface. The implementation results demonstrate that the system is capable of performing near real-time skin disease classification on mobile devices.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Aplikasi mobile, Convolutional Neural Network, FastAPI, Flutter, Klasifikasi penyakit kulit, Mobile application, MobileNetV2, Skin disease classification
Subjects:	T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Zakia Kolbi
Date Deposited:	23 Jan 2026 06:12
Last Modified:	23 Jan 2026 06:12
URI:	http://repository.its.ac.id/id/eprint/130195

Actions (login required)

View Item