Plant diseases are a leading cause of loss of food security worldwide and thus the management of diseases at the initial level is very important. Current identification approaches involve the use of manual inspection which is labor intensive and can be inaccurate or the traditional machine learning which has limitations of a lot of feature engineering. The proposed Plant Disease Identification System solves these problems by using Convolutional Neural Networks (CNNs) which are efficient in feature extraction and image classification. The system increases the accuracy of detection in a three stage process; the first stage is image acquisition, the second stage is image preprocessing which aims at removing noise and enhancing features and the last stage is classification where the image is classified using a CNN that has been trained on a large database of images of diseased plants. As compared to the existing techniques, this system provides more accurate and efficient results and is a convenient tool which can be used by farmers and agricultural experts to control the health of crops.