A Comparative Survey on Paddy Leaf Diseases Detection using Convolutional Neural Network Algorithms

Despite recent difficulties in rice cultivation, rice leaf diseases are still identified using traditional methods. This leads to a fall in production of the crops. Initial detection allows farmers to increase output rates and efficiently take the necessary treatment measures. Convolutional Neural Network (CNN) algorithms’ quick development has significantly modified agricultural practices, particularly in disease identification. This paper comprehensively examines the use of CNN-based approaches for identifying illnesses in rice leaves. Given the rising need for rice production and quality on a global scale, accurate and quick disease diagnosis is essential. This review study also deals with the different CNN architectures and the corresponding modifications to detect rice disease. It also includes single - and multi-class classification algorithms. In addition, examining important datasets, preprocessing methods, and augmentation strategies reveals their impact on the performance of the developed models. Comparative evaluations of the surveyed papers highlight the potential and constraints of various CNN models and suggest fresh directions for investigations. This study highlights the potential of CNNs to completely reinvent across automated and effective verification, which ultimately increases agricultural productivity.