Pengembangan Model Hybrid DenseNet-SVM Untuk Klasifikasi Penyakit Buah Jambu Berdasarkan Citra Digital

Reza Novriansah (Universitas Bina Insan)
Asep Toyib Hidayat (Universitas Bina Insan Lubuklinggau)
Harma Oktavia LW3 (Universitas Bina Insan)

Jurnal Teknologi Informasi MURA
Vol. 18 No. 1 (2026): Jurnal Teknologi Informasi Mura MARET

Abstract Diseases in guava fruit such as Anthracnose and fruit fly attacks can drastically reduce crop quality. Manual identification is often subjective and slow. This research proposes a hybrid model combining Deep Learning architecture DenseNet121 as a feature extractor and Support Vector Machine (SVM) as a classifier. The dataset used is the "Guava Disease Dataset" which has been augmented into 3,784 images. The results showed that the hybrid DenseNet-SVM model with a linear kernel achieved the highest testing accuracy of 99.62%. This proves that combining deep feature extraction with an optimal margin classifier is highly effective for plant disease detection. Keywords— guava, disease classification, DenseNet121, SVM, digital image