Cotton Diseases Multiclassification: CNN and Random Forest Approach

The study of cotton diseases aims to reduce their destructive effect on agriculture, standard methods for identifying diseases, such as observation and specialized diagnosis, are costly, slow, are subject to human error. This paper uses the hybrid approach and data sizes. Convolutional neural networks are now a type of artificial intelligence model composed of a layer and a pooling layer. This feature enhances its effectiveness. Since deep learning, as well as neural networks using convolution in particular, are capable of capturing intricate visual structures while accomplishing superior classification precision, they have demonstrated immense potential in recognizing of cotton diseases. This update made reference to a wide range of measures, namely average, preciseness, support, and recall. The minimum accuracy in Class 6 is 0.96 and the maximum accuracy in Class 4 is 0.98. The macro average for precision is 88.17, recall is 90.36 and the F1-Score is 89.25 respectively. Micro Average for precision is 89.64, recall is 92.89 and F1-Score is 91.23 respectively. The Macro Average for precision, recall, and F1-Score is 91.06, 90.99, and 91.00 respectively. In this, it is used 8 classes for classification. It is helpful for farmers because it saves money and time.