An Efficient Plant Disease Classification Using Pre-trained CNN Model and Handcrafted Features

The amount and aggregate of fruits are affected by various diseases, therein disturbing the economy of a country. Some issues such as immaterial features and the vector dimensionality increases the estimated time of the disease detection system. So, an integrated deep learning framework is proposed in this work to classify the tomato plant diseases. The proposed method employs handcraft feature extraction along with pre-trained deep learning features. Then, these features are merged together using feature fusion technique. Later on, the resultant fused vector undergoes a feature selection algorithm. Finally, support vector machine is used to classify the different disease present in the tomato dataset. The efficacy of this proposed work is evaluated on the Plant-Doc dataset, and an accuracy of 98.8 is achieved. The experimental analysis shows that the proposed method shows better performance over other state-of-the-art tomato disease detection techniques previously reported in the literature.