Transfer Learning based Rice Leaf Disease Classification with Inception-V3
Kotharu uma Venkata Ravi teja, B Pavan Venkat Reddy, Likhitha Reddy Kesara, Kotaru Drona Phani Kowshik, Lakshmi Anchitha Panchaparvala
Abstract
In the agriculture field, the detection of diseases from various plants through images is one of the most vital areas that needs to be ameliorated. The collection of cultivating crops and alleviation of features are the source of plant infection. A range of spot evaluation methods and disease diagnoses have been applied and advanced in a broad variety of crops. The Deep Learning concept of artificial intelligence is applied. However, this article introduces a model to detect disease in less time. Image processing techniques are used for operating the data with various methods. Three distinct rice plant diseases are included in the dataset: brown spot, leaf smut, and bacterial leaf blight. Spotting the disease on plants has been performed with various techniques like transfer learning and image data generators. In transfer learning, there are four types of approaches used: reusing the trained model, pre-trained model, feature extraction, and popular trained models. A popular model used to train the data is InceptionV3. Categorical cross-entropy is used for calculating the loss and optimization of the model. A surpassing result of 99.33% accuracy has been achieved on the testing dataset.