Design and development of a robust Deep learning model for detection of disease cotton crop

Abstract

A deep learning model to identify damage to cotton leaves is proposed in this work, which is based on photographs of cotton harvesting in the field. Deep learning models are used to identify damage to cotton leaves. Cotton is one of the most profitable horticultural crops in the world, and it has been cultivated in virtually every part of the planet. It has become the focus of a wide range of farming nuisances and diseases in tropical climates, prompting the deployment of efficient control methods in these regions. Additionally, it is difficult for the creator to distinguish between the affects of the primary nuisances and diseases in the underlying stages, making it tough for them to locate the proper identifying proof of a harm in the underlying stages. The current work gives an answer based on significant expertise in the screening of cotton leaves, which allows for improved monitoring of the health of the cotton crop and the ability to make better judgments regarding its management. With the use of convolutional neural networks, two convolutional brain models — GoogleNet and Resnet50 — achieved accuracy rates of 86.6 percent and 89.2 percent, respectively, in their classifications. This approach outperforms standard methods for managing pictures, such as support vector machines (SVM), closest k-neighbors (KNN), imitating brain organisations, and neuro-fluffy, by a factor of 25. (NFC). This means that this strategy may be able to contribute to a more speedy and accurate analysis of the plants that are populating the area in the future.