The agricultural production of the country is severely affected when herbs and crops are attacked by disease. The usual methods adopted by farmers or even agriculture experts are to make several observation to the herbs with naked eye in order to identifying and detecting the disease, and make an approximate decision for herbs treatment. This method happens to be always a time consuming and inaccurate that leads to be expensive. Now we have advanced technology such as automatic detection using deep learning, which produce results accurate and fast. This paper aims to present an approach to develop a model to detect herbs disease progress, depending on the leaf images classification, using deep convolutional network. With the advent of computer vision, it has been noticed that the precision herbal protection were improvised and therefore the computer vision applications have gained more popularity even in precision agriculture field. Here, Novel training techniques are proposed which actually enables faster and less complex implementations in order to herb dieses detection. All the necessary key steps required in order to implement disease detection model has been described in this paper. These key steps ranges from collection of images to building database, evaluated by experts in the field of agriculture with the assistance of deep CNN training are described. The described technique is nothing but intelligent system development in order to classifying herb infections by means of deep convolutional neural networks. This model were trained and tweaked in order to suit the database of herb’s leaves images, that were congregated self-sufficient for diverse plant diseases. The growth and novelty of this developed model dwell in its simplicity. Healthy leaves and background images match other classes, allowing the model to use CNN to distinguish between diseased leaves and healthy leaves.