Plant disease monitoring and management are essential for ensuring reliable and lucrative crop production in all kinds of plantations and guaranteeing sustainable agriculture production. Most traditional approaches depend significantly on human effort, which is liable to time delay and is expensive. Moreover, plant pathogens are nearly identical to other non-harmful species in many circumstances. Recently, computer vision based deep learning algorithms have not been deceived by these similar diseases causing false warnings. Thispaper proposes a novel deep ensemble neural network (D-ENN) framework for automated plant disease detection. The dataset collected in real cultivated fields contains healthy and diseased images with specific class labels. Since there are limited images of a few specific crops, a conditional generative adversarial network (CGAN) is leveraged to generate the additional synthetic images. Then, the total dataset is split into the training set, validation set, and test set, in the ratio of 70:10:20 used to avoid overfitting problems. The proposed model istrained using real and synthetic images utilizing the transfer learning mechanism. Finally, the experimental outcomes are assessed using standard performance measures evaluating the performance of the proposed method. The proposed vCrop framework attained Precision, Recall, and F1-Measure, and Accuracy of 95.71%, 95.32%, 95.51%, and 96.02% respectively, in classifying the plant diseases in comparison with the other stateof- the-art approaches. The proposed D-ENN model can be a potentially helpful tool for farmers and agronomists in diagnosing and quantifying cotton diseases.