Abstract
In deep learning perspective, convolutional neural network (CNN) forms the backbone of image processing. For reducing the drawbacks and also to get better performance than conventional neural network, the new architecture of CNN known as, capsulenet is implemented. In this paper, we analyze capsulenet for two datasets, which are based on plants. In the modern world, most of the diseases are contaminating due to the lack of hygienic food. One of the main reasons for this is, diseases affecting crop species. So, the first model is built for plant disease diagnosis using the images of plant leaves. The corresponding dataset consists of 54,306 images of 14 plant species. The proposed architecture with capsulenet gives an accuracy around 94%. The second task is plant leaves classification. This dataset consists of 2,997 images of 11 plants. The prediction model with capsulenet gives an accuracy around 85%. In the recent years, the use of mobile phones is rapidly increasing. Here for both the models, the images of plant leaves are taken using mobile phone cameras. So, this method can be extended to various plants and can be adopted in large scale manner.