Most illnesses of plants have obvious signs, as well as the current acknowledged method involves having a skilled plant pathologist visually inspect the diseased leaves of the plant to arrive at the conclusion. Computer-aided diagnostic techniques are well suited to this task because the illness diagnosis procedure is laborious to complete manually and because the pathologist’s skill level directly affects the diagnosis’s accuracy. We need a model that can conduct successful classification without pre-processing, as contrast to standard machine learning techniques that demand flawless manual extraction of features to yield successful results. This paper suggested the Efficient Net deep studying architecture in order to categorize plants leaf diseases and also compared its performance toot her cutting edge deep learning models. To train models, the Plantation Kingdom information was employed. Unique and enhanced facts with 55,448 and 61,486 photos were employed for instruction each one, correspondingly. Transfer gaining knowledge was employed for training the Efficient Net additional deeper learning algorithms as well as framework. Every model layer in the transfer learning was set up to allow for training. The test dataset’s findings demonstrated that the Efficient Net architecture’s B5 and B4 models outperformed other deep learning models with regard to precision and precision, scoring 99.90% and 99.96%, respectively, and 98.49% and 99.49%, respectively, in the original and supplemented datasets.