| Sumario: | Rice is central to the lives of billions of people around the world. It is affected by different diseases at all growth periods of its cultivation. In plants, diseases are predominantly caused by biotic and abiotic components such as fungi, viroids, nematodes, bacteria, viruses, temperature, nutrient deficiencies and other environmental conditions. The important diseases of rice such as leaf blast, bacterial blight, and brown spot cause damage to rice can greatly reduce yield. Farmers globally deal with the problem of plant diseases diagnosis and for their appropriate treatment. Therefore, real-time and precision identification of rice leaf diseases is urgently needed. Recent developments in DL (Deep Learning) approaches have tremendously increased the capabilities of visual recognition systems through computer vision technologies. The most popular DL models leverage for computer vision problems are convolutional neural networks (CNNs) that has confirmed very successful in areas such as image categorization, object detection, image segmentation, etc. It has paved the way for automatic plant disease detection using plant images. In this research, the dataset is limited to train a deep convolutional neural network models such as VGG-16, ResNet50 and InceptionV3 in this case transfer learning came into picture. Transfer learning is a popular technique in deep learning where pre-trained models are re-purposed on a related problem. The main purpose of this study is to assess deep convolutional neural networks with transfer learning for the identification of different diseases in the rice plant leaf. The proposed CNNs were utilized with transfer learning technique to lessen the training time and enhance the functional capabilities of neural networks. The classification accuracies for the VGG-16, ResNet50 and InceptionV3 CNN models were 87%, 93%and 95% respectively.
|
|---|