Plant diseases act as a major threat to the both economy and food security of
any nation. Despite being of such importance, the identification of plant diseases and
approaches deployed to tackle them are mostly conventional/ traditional ones.
Incubation of technology and advancement in computer vision and deep learning
models have opened new ways for developing much better approaches to tackle such
issues. In this work, the native plants of Jammu and Kashmir are taken into
consideration. An IoT-based framework is designed for data collection and disease
diagnosis. The data involves both diseased and healthy leaf images. A hybrid deep
neural network is trained to identify the plant species as well as the diseases associated
with it. The trained model achieves an overall accuracy of 96.35%. A comparison with
other state of art approaches is also presented, along with suggestions for some related
future developments. This approach can be deployed on a global scale to tackle plant
diseases and to achieve global diagnosis.
Keywords: Computer vision (CV), Internet of things (IoT), Machine learning (ML), Neural networks (NN), Precision agriculture (PA), Rarefied flow.
