<![CDATA[Abstrak Perkembangan teknologi kecerdasan buatan (Artificial Intelligence/AI) memberikan dampak signifikan dalam bidang pertanian, khususnya pada deteksi dan klasifikasi penyakit tanaman. Penelitian ini mengusulkan model hibrid yang mengintegrasikan Residual Network (ResNet) sebagai ekstraktor fitur dengan Support Vector Machine (SVM) sebagai classifier utama untuk mengklasifikasikan penyakit pada tanaman jagung berbasis citra daun. Dataset yang digunakan mencakup empat kelas, yaitu Blight, Common Rust, Gray Leaf Spot, serta daun jagung Healthy atau sehat. Hasil pengujian menunjukkan bahwa model hibrid ResNet-SVM mampu mencapai akurasi akhir sebesar 94,61%. Berdasarkan laporan klasifikasi, performa terbaik ditunjukkan pada kelas Healthy dengan nilai precision, recall, dan f1-score mencapai 1,00. Kelas Common Rust juga memperoleh hasil tinggi dengan f1-score 0,96, sedangkan kelas Blight mencapai f1-score 0,92. Namun, kelas Gray Leaf Spot masih menjadi tantangan dengan f1-score 0,62 akibat jumlah data yang relatif lebih sedikit. Secara keseluruhan, nilai macro average f1-score tercatat sebesar 0,88, sementara weighted average f1-score mencapai 0,94. Temuan ini menunjukkan bahwa kombinasi ResNet dan SVM efektif dalam meningkatkan akurasi klasifikasi penyakit jagung, sekaligus memperkuat potensi penerapan metode hibrid deep learning dan machine learning dalam sistem deteksi penyakit tanaman berbasis citra digital. Kata kunci: Resnet, SVM, Model Hibrid, Klasifikasi, Penyakit Jagung Abstract The advancement of Artificial Intelligence (AI) has significantly impacted agriculture, particularly in plant disease detection and classification. This study proposes a hybrid model that integrates Residual Network (ResNet) as a feature extractor with Support Vector Machine (SVM) as the main classifier for classifying corn leaf diseases based on image data. The dataset consists of four classes: Blight, Common Rust, Gray Leaf Spot, and Healthy leaves. Experimental results show that the hybrid ResNet-SVM model achieved a final accuracy of 94.61%. The best performance was obtained in the Healthy class with precision, recall, and f1-score of 1.00. The Common Rust class also achieved a high f1-score of 0.96, while the Blight class reached 0.92. However, the Gray Leaf Spot class remained more challenging, with an f1-score of 0.62 due to the relatively smaller number of samples. Overall, the model achieved a macro average f1-score of 0.88 and a weighted average f1-score of 0.94. These findings demonstrate that the combination of ResNet and SVM is effective in enhancing classification accuracy compared to single methods, highlighting its potential application in developing automated corn disease detection systems based on digital leaf images. Keywords: ResNet, SVM, hybrid model, classification, corn disease]]>
Copyrights © 2025
