The presence of plant diseases significantly impacts the overall productivity and quality of crops. An early identification of plant disease will help farmers for improving the overall crop yield production. The deep neural network methods have been recently applied in these kinds of problems for getting accurate results. This research work implemented fully convolutional DenseNet for predicting diseases from plant images. The rice leaf images are first collected from Kaggle and converted as a greyscale colour format. These converted images processed using the encoding and decoding layers of fully convolutional DenseNet architecture. The function of Softmax activation is applied on the final fully connected layer for identifying the disease class labels in rice leaf images. The effectiveness of the proposed plant disease prediction algorithm is assessed through metrics such as accuracy, precision, recall, and F1-score, and is subsequently compared with current state-of-the-art plant disease classification techniques.