Plant diseases destroy crops, costing the agricultural business money. Although pesticides have been used to boost agricultural output, their abuse is harmful to the environment. As a consequence, the capacity to diagnose illnesses and differentiate them from nutritional deficits has a significant impact on assessing whether pesticides are essential. Traditional approaches for diagnosing plant diseases in the lab entail time-consuming and difficult chemical procedures. This work proposes an automated strategy that combines machine learning (ML) and image processing techniques to detect and categorise plant illnesses. To train the algorithm on photos, the feature extraction approach is utilised. In order to choose the optimal algorithm for illness detection, the efficacy of multiple deep learning algorithms is tested using training information. The unseen images may be found in the test folder, which is meant to put the system's capacity to detect plant ailments to the test. The overall accuracy of the procedure is 95%. A vast number of photos may be utilised to train the system, resulting in faster and more accurate outputs.