Paddy cultivation is facing significant challenges due to various types of diseases that can severely impact crop yield and the product quality. This research work uses a publicly available dataset for paddy diseases comprising of images for different types of paddy diseases and healthy paddy plants. The recent advancements in the field of deep learning, neural network architectures, and image processing techniques enhance the power of detecting discriminative features from paddy leaf images automatically. The research focusses on pre-trained deep learning models VGG19, ResNet50, InceptionV3, MobileNet with transfer learning for paddy disease detection. The workflow involves preprocessing steps such as resizing, scaling, shearing, and normalization. Then the preprocessed images are classified in the above models using transfer learning approach. The efficiency of models are evaluated in terms of accuracy, precision, F1-score, and recall of the deep learning-based approach with transfer learning techniques in accurately classifying paddy diseases. The experimental results show that better results are achieved with ResNet50, classifying paddy plant diseases with an impressive accuracy of 99.98% while InceptionV3 recorded 83.07%, MobileNet recorded 99.02%, and VGG19 recorded as 90.14%. The proposed system helps farmers as a reliable tool for monitoring and reducing the effects of various paddy diseases, ultimately contributing to sustainable agriculture practices and food security.