Wheat Disease Severity Assessment Using Federated Learning CNNs for Agriculture Transformation
H.K. Narang, Rishabh Saklani, Karan Purohit, Purushottam Das, Bharat Bhushan Sagar, Manisha Manjul
Abstract
In order to stop the spread and improve the agricultural productivity it is important to rightly detect and assess infection of wheat. The research's goal is to assess the seriousness of wheat diseases using CNN and collaborative learning. In this plants of wheat having 5 different diseases namely leaf rust, fusarium head blight, wheat stripe mosaic virus, powdery mildew and septoria leaf blotch are photographed. The model has been trained via federated learning that enables several devices with the capability to simultaneously train the machine learning model while maintaining the data privacy. The results displayed that CNN with federated learning is capable of predicting the seriousness of various wheat illnesses accurately with an average accuracy of 0.85. This enables the farmers and specialists to take appropriate measures to avoid disease spread and boost farming productivity. The results of the study demonstrate how cooperative learning works. CNN is an option to improve agricultural productivity, halt the spread of infections, and assess the seriousness of wheat problems.