Ricky Eka Putra, Anita Qoiriah, Guntur Trimulyono, Dian Candra Rini Novitasari, Dina Zatusiva Haq, Rinda Nariswari
Corn plays a central role as a staple food with a significant impact on the global economy, environment, and nutrition. Apart from providing nutrition for humans, the high demand for corn is influenced by its use in global animal feed and ethanol production. Corn plants rank third globally in planted area, making it challenging to maintain corn quality due to leaf damage. Therefore, the development of an efficient leaf-damage identification system for corn plants is critical for monitoring and improving production quality. This research combines the Support Vector Machine-Recursive Feature Elimination (SVM-RFE) method with GoogLeNet's Convolutional Neural Network (CNN) architecture after feature learning. SVM-RFE contributes to feature selection by reducing the feature dimensions of the GoogLeNet CNN extraction results, focusing on the most relevant features for the classification of leaf diseases in corn plants. The adaptive ability of SVM to remove low-impact features is strengthened by GoogLeNet CNN's ability to extract complex patterns from images. Experimental results show that the GoogLeNet-SVM-RFE method with 235 features achieves the highest accuracy, sensitivity, and specificity of 99.28%, 99.25%, and 98.93%, respectively, using only 23% of the original features, indicating high efficiency. With excellent performance, the GoogLeNet-SVM-RFE method with 235 features takes only 41 seconds, demonstrating significant superiority over conventional approaches in both training time and accuracy. © 2026, Politeknik Negeri Padang. All rights reserved.
Department of Informatics, Faculty of Engineering, Universitas Negeri Surabaya, Surabaya, Indonesia; Undergraduate Program of Biology, Faculty of Mathematics and Natural Sciences, Universitas Negeri Surabaya, Surabaya, Indonesia; Mathematics Department, Faculty of Science and Technology, UIN Sunan Ampel Surabaya, Surabaya, Indonesia; Department of Informatics, Faculty of Computer Science, Universitas Pembangunan Nasional Veteran Jawa Timur, Surabaya, Indonesia; Statistics Department, School of Computer Science, Bina Nusantara University, Jakarta, Indonesia
