Cognitive Analytics and Reinforcement Learning

11A Lightweight CNN Architecture for Prediction of Plant Diseases

Sasikumar A.¹, Logesh Ravi², Malathi Devarajan³, Selvalakshmi A.⁴ and Subramaniyaswamy V.⁵*

¹Department of Data Science and Business Systems, Faculty of Engineering and Technology, SRM Institute of Science and Technology, Kattankulathur, Tamil Nadu, India

²Centre for Advanced Data Science, Vellore Institute of Technology, Chennai, India

³School of Computer Science and Engineering, Vellore Institute of Technology, Chennai, India

⁴Vellore Institute of Technology, Chennai, India

⁵School of Computing, SASTRA Deemed University, Thanjavur, India

Abstract

Detecting maize crop diseases accurately is a difficult task that farmers face throughout the maize development and production stages. TensorFlow will be used for Deep Learning and the Convolutional Neural Network (CNN) architecture. Existing CNN-based solutions have been demonstrated to be less accurate and incompatible with most datasets and applications. We developed a real-time, automated system for detecting plant illnesses based on the CNN architecture. Using a reinforcement learning model on synthetic and real-world images, the proposed system is trained to categorize the input into sickness. The proposed architecture provides high accuracy, a lightweight model, and speedy processing. The experimental result shows that the proposed architecture accurately predicted types of plant diseases, and the system performance did not decline over time. It can be used ...