Plant Leaf Perception Using Convolutional Neural Network

Automatic plant leaf disease prediction is a complex and essential research topic-The main aim of plant leaf perception using CNN is to detect diseases that occur on different plant leaves, the proposed system used deep learning methods to detect the various diseases on the leaves of such plants. Deep learning architecture selection was the key issue for the implementation. In earlier stages disease in plants was predicted by using some classic image processing techniques like threshold, contrast enhancement and morphological contour operations, then to make disease detection in advance level they have used data mining applications such as classification and clustering approaches for predicting infected leaf diseases. But in both the above mentioned works there are some impacts in processing the images, it requires manual disease prediction, then time complexity of work was high and the image with high qualities will have different accuracies and low quality image can’t be accessed for processing. Considering all these limitations, the proposed work is implemented, we have used Convolutional Neural Network (CNN) algorithm for automatic prediction of plant leaf disease. Here, the proposed work has divided into four stages, such as data collecting, data wrangling (preprocessing the collected data) i.e, unwanted data will be removed. After analyzing the dataset training and testing the model by using collected data will be done. Finally, the CNN algorithm to classify the plant leaf image and identify the disease information is applied. Thus, this project results with high accuracy of 93% (approximately) and effective prediction of plant disease with minimum time.