Repository Universitas PakuanPEMODELAN KLASIFIKASI PENYAKIT PADA DAUN ANGGUR MENGGUNAKAN CONVOLUTIONAL NEURAL NETWORK (CNN)Novia PermataSariauthorPEMODELAN KLASIFIKASI PENYAKIT PADA DAUN ANGGUR MENGGUNAKAN CONVOLUTIONAL NEURAL NETWORK (CNN) Novia Permata Sari1 ; Sufiatul Maryana2 ; Adriana Sari Aryani3 . Program Studi Ilmu Komputer1,2,3 Universitas Pakuan, Bogor, Indonesia1,2,3 www.unpak.ac.id1,2,3 Nopiapermatasar518@gmail.com1 , sufiatul.maryana@unpak.ac.id2 , adriana.aryani@unpak.ac.id3 (*) Corresponding Author The creation is distributed under the Creative Commons Attribution-NonCommercial 4.0 International Lincense. Abstract- Grapevine diseases often pose a major challenge, leading to reduced crop quality and yield if not detected early. Manual disease identification typically requires specialized expertise, high precision, and considerable time. This study aims to develop an automated deep learning-based classification model for grapevine leaf diseases using the Convolutional Neural Network (CNN) method with a DenseNet121 base architecture. The dataset consists of 1,600 digital images of green grapevine leaves (*Vitis vinifera* L. var. Thompson Seedless) collected from plantations in Sukaraja District, Bogor Regency. These images are evenly distributed across four condition classes: Healthy, Black Measles (Esca), Leaf Blight (Isariopsis Leaf Spot), and Black Rot. Model testing was conducted using two data split scenarios: 80:10:10 and 70:15:15. The results demonstrate that the DenseNet121 architecture effectively extracts visual image characteristics. The best performance was achieved with the 70:15:15 data split, yielding a validation accuracy of 91.25%, whereas the 80:10:10 split resulted in an accuracy of 88.75%. The Healthy and Leaf Blight classes were identified with 100% accuracy. This classification model was integrated into a web application built on the Flask framework as a practical implementation to support smart a validation accuracy of 91.25%, whereas the 80:10:10 split resulted in an accuracy of 88.75%. The Healthy and Leaf Blight classes were identified with 100% accuracy. This classification model was integrated into a web application built on the Flask framework as a practical implementation to support smart agriculture systems. Keywords: Convolutional Neural Network (CNN), Deep Learning, Image Classification, Grape Leaf Disease, DenseNet121. Abstrak- Penyakit pada tanaman anggur sering menjadi kendala utama yang menyebabkan penurunan kualitas dan kuantitas hasil panen apabila tidak dideteksi sejak dini. Identifikasi penyakit secara manual umumnya membutuhkan keahlian khusus, ketelitian tinggi, dan waktu yang relatif lama. Penelitian ini bertujuan untuk membangun model klasifikasi otomatis penyakit pada daun anggur berbasis Deep Learning menggunakan metode Convolutional Neural Network (CNN) dengan arsitektur dasar DenseNet121. Dataset yang digunakan berupa 1.600 citra digital daun anggur hijau (Vitis vinifera L. var. Thompson Seedless) yang dikumpulkan dari perkebunan di Kecamatan Sukaraja, Kabupaten Bogor. Citra tersebut terbagi rata ke dalam empat kelas kondisi: Healthy (Sehat), Black Measles (Esca), Leaf Blight (Isariopsis Leaf Spot), dan Black Rot. Pengujian model dilakukan menggunakan dua skenario pembagian proporsi data, yaitu 80:10:10 dan 70:15:15. Hasil penelitian menunjukkan bahwa arsitektur DenseNet121 mampu mengekstraksi karakteristik visual citra secara optimal. Performa terbaik dicapai pada skenario proporsi data 70:15:15 dengan tingkat akurasi validasi sebesar 91,25% , sedangkan proporsi 80:10:10 menghasilkan akurasi sebesar 88,75%. Kelas Healthy dan Leaf Blight berhasil diidentifikasi secara sempurna dengan akurasi 100%. Model klasifikasi ini diintegrasikan ke dalam aplikasi web berbasis framework Flask sebagai implementasi praktis pendukung sistem pertanian cerdas (smart agriculture).Ilmu Komputer2026-06-21Universitas Pakuan;Fakultas Matematika dan Pengetahuan AlamThesis

For work being deposited by its own author: In self-archiving this collection of files and associated bibliographic metadata, I grant Repository Universitas Pakuan the right to store them and to make them permanently available publicly for free on-line. I declare that this material is my own intellectual property and I understand that Repository Universitas Pakuan does not assume any responsibility if there is any breach of copyright in distributing these files or metadata. (All authors are urged to prominently assert their copyright on the title page of their work.)

For work being deposited by someone other than its author: I hereby declare that the collection of files and associated bibliographic metadata that I am archiving at Repository Universitas Pakuan) is in the public domain. If this is not the case, I accept full responsibility for any breach of copyright that distributing these files or metadata may entail.

Clicking on the deposit button indicates your agreement to these terms.