Novel predictive analysis for leaf diseases using convolutional neural networks comparing over artificial neural networks.

To enhance the detection of leaf disease through image recognition and classifying leaf diseases using Convolutional neural networks and to compare the performances in leaf disease through novel predictive analysis with artificial neural networks. Convolutional neural networks find the optimal policy by learning the optimal values for each state-action pair. The implementation has been carried out for novel predictive analysis with the help of python libraries such as Tensorflow and Keras. The algorithms tested over 45 sets of early blight potato leaves and late blight potato leaves images and 45 sets of healthy potato leaves images which have been utilized for test and training dataset images respectively, that are classified for recognition of diseases. The novel predictive analysis on the data set and test cases has been performed successfully and acquired 91% accuracy for the Convolutional neural networks algorithm and compared to the artificial neural network algorithm (ANN), which gave 80% accuracy. With the level of significance (p<0.005), the resultant data depicts the reliability in independent sample t tests. On the whole process of prediction of accuracy for the image recognition, the Convolutional neural networks algorithm gives significantly better performance when compared to ANN algorithm. [ABSTRACT FROM AUTHOR]