Automatic Tomato Leaf Diseases Classification and Recognition using Transfer Learning Model with Image Processing Techniques

In agriculture, manual approaches to recognize tomato plant diseases are often time-consuming. The diseases spread throughout the plant, affecting the quality and yield. In this research work, a transfer learning model (DenseNet201) was designed to conduct the symptom-wise recognition of three tomato diseases such as leaf mold, early blight, and late blight are compared with healthy leaves. The transfer learning model has been trained, tested, and validated. Our trained model can recognize the category of disease. Once the images are classified, the regions are segmented, and features are extracted using k-means clustering and ORB keypoints. The transfer learning model achieved an excellent recognition result with an average accuracy of 99.688%.