Pertanian padi merupakan sektor penting dalam menjamin ketahanan pangan, namun sering menghadapi tantangan berupa penyakit daun yang secara signifikan mengurangi hasil panen. Penelitian ini mengembangkan sebuah aplikasi untuk mendeteksi penyakit daun padi menggunakan algoritma Convolutional Neural Network (CNN), dengan tujuan meningkatkan efisiensi, produktivitas, dan memberdayakan petani melalui teknologi digital. Metodologi penelitian melibatkan pengumpulan data melalui observasi di Klinik PHT Binakarya III, wawancara, tinjauan pustaka, dan pengembangan sistem menggunakan metode Rapid Application Development (RAD). Evaluasi model dilakukan dengan 120 gambar uji dan mencapai akurasi keseluruhan sebesar 91,6%, dengan F1-Score masing-masing 0,921 untuk Blas, 0,871 untuk Hispa, dan 0,952 untuk Sehat (daun sehat). Untuk meningkatkan kinerja model, perbaikan di masa depan meliputi peningkatan ukuran dataset, penerapan teknik augmentasi data yang lebih beragam, analisis mendalam terhadap gambar-gambar yang sulit, serta penambahan klasifikasi untuk mencakup lebih banyak jenis penyakit. Penelitian ini menunjukkan potensi signifikan dari aplikasi yang diusulkan dalam mendukung deteksi dini penyakit daun padi, sehingga berkontribusi pada praktik pertanian berkelanjutan dan manajemen tanaman yang lebih baik.

Rice farming is a vital sector for ensuring food security but often faces challenges from leaf diseases that significantly reduce crop yields. This study developed an application to detect rice leaf diseases using a Convolutional Neural Network (CNN) algorithm, aiming to improve efficiency, productivity, and empower farmers through digital technology. The methodology involved data collection through observations at the PHT Binakarya III Clinic, interviews, literature reviews, and system development using the Rapid Application Development (RAD) method. Model evaluation conducted with 120 test images achieved an overall accuracy of 91.6%, with F1 scores of 0.921 for Blas, 0.871 for Hispa, and 0.952 for Sehat (healthy leaves). To enhance the model's performance, future improvements include increasing the dataset size, applying diverse data augmentation techniques, conducting detailed analyses of challenging images, and expanding the classification to include a broader range of diseases. This study demonstrates the significant potential of the proposed application in supporting the early detection of rice leaf diseases, thereby contributing to sustainable farming practices and improved crop management.

Rice farming is a vital sector for ensuring food security but often faces challenges from leaf diseases that significantly reduce crop yields. This study developed an application to detect rice leaf diseases using a Convolutional Neural Network (CNN) algorithm, aiming to improve efficiency, productivity, and empower farmers through digital technology. The methodology involved data collection through observations at the PHT Binakarya III Clinic, interviews, literature reviews, and system development using the Rapid Application Development (RAD) method. Model evaluation conducted with 120 test images achieved an overall accuracy of 91.6%, with F1 scores of 0.921 for Blas, 0.871 for Hispa, and 0.952 for Sehat (healthy leaves). To enhance the model's performance, future improvements include increasing the dataset size, applying diverse data augmentation techniques, conducting detailed analyses of challenging images, and expanding the classification to include a broader range of diseases. This study demonstrates the significant potential of the proposed application in supporting the early detection of rice leaf diseases, thereby contributing to sustainable farming practices and improved crop management.