Most of the crops which are grown in India have shown reasonable production from past decades. Among this paddy is considered as the most pre-dominant crop of our country. Mostly the crops are subject to common leaf diseases such as brown spot, bacterial leaf blight, leaf blast and tungro. These diseases were reported from almost more than 70 countries facing the same diseases. It is necessary to monitor the growth of the paddy crops and detect the disease with relevant measures to save the crop. The current research work gives a key solution using novel deep learning techniques to identify and differentiate between diseases at earlier stage. A novel method is proposed by tuning densenet CNN architecture with a smaller number of parameters. The proposed method is compared with the state-of-art methods such as VGG-Net, Mobile Net, Res-Net and the fine-tuned dense model achieves 98% accuracy and other performance measures also satisfied. Also, the proposed method achieves sensible computational time.