Tanaman tomat (Solanum lycopersicum) merupakan komoditas hortikultura bernilai ekonomi tinggi yang rentan terhadap serangan penyakit daun, seperti Bercak Daun (Leaf Spot), Busuk Daun (Leaf Blight), dan Jamur Daun (Powdery Mildew). Kemiripan gejala visual antarpenyakit menyebabkan kesalahan identifikasi manual yang berdampak pada penggunaan pestisida tidak tepat dan penurunan hasil panen petani di Desa Babakanmulya, Kecamatan Cigugur, Kabupaten Kuningan. Penelitian ini bertujuan merancang dan mengimplementasikan aplikasi klasifikasi penyakit daun tomat berbasis Android menggunakan algoritma Convolutional Neural Network (CNN) dengan arsitektur MobileNetV2. Metode pengembangan yang digunakan adalah Rapid Application Development (RAD), dengan data dikumpulkan melalui observasi, wawancara, kuesioner, dan studi pustaka. Dataset terdiri dari 4.000 citra empat kelas dengan proporsi 80% pelatihan, 10% validasi, dan 10% pengujian. Model dibangun melalui transfer learning berbasis MobileNetV2, dilatih selama 100 epoch, dan dikonversi ke TensorFlow Lite untuk perangkat Android. Hasil pengujian menunjukkan seluruh skenario black box testing valid, white box testing menghasilkan Cyclomatic Complexity bernilai 2 tanpa kesalahan logika, dan User Acceptance Test (UAT) mencapai tingkat kelayakan 91,07%. Aplikasi terbukti layak digunakan sebagai alat bantu identifikasi penyakit daun tomat secara otomatis, cepat, dan akurat.
tomat secTomato (Solanum lycopersicum) is a high-value horticultural commodity highly susceptible to leaf diseases, including Leaf Spot, Leaf Blight, and Powdery Mildew. The visual similarity among disease symptoms leads to misidentification through manual inspection, resulting in improper pesticide use and reduced crop yields among farmers in Babakanmulya Village, Cigugur District, Kuningan Regency. This study aims to design and implement an Android-based tomato leaf disease classification application using the Convolutional Neural Network (CNN) algorithm with MobileNetV2 architecture. The Rapid Application Development (RAD) method was applied, with data collected via observation, interviews, questionnaires, and literature review. The dataset comprised 4,000 images across four classes, split into 80% Training, 10% validation, and 10% testing. The model was built using transfer learning on pre-trained MobileNetV2, trained for 100 epochs, and converted to TensorFlow Lite for on-device inference. Black box testing confirmed all functional scenarios as valid; white box testing yielded a Cyclomatic Complexity of 2 with no logical errors; and the User Acceptance Test (UAT) achieved an acceptability score of 91,07%. The application is proven to be an effective and feasible tool for automated tomato leaf disease identification.ara otomatis, cepat, dan akurat.