Plant diseases are undesirable conditions that significantly reduce crop growth and quantity. Observing plants for disease with the naked eye is a common practice among expert biologists and farmers, although it can be inaccurate and take a lot of time. In this research, we build and construct an intelligent classification method for leaf diseases using computer vision and artificial intelligence approaches. In this study, two approaches are used, and the results of their simulations are analyzed to evaluate performance. Convolutional neural networks are utilized to extract the deep properties of the plants from the photos of the plants from the rice, potato, and tomato Plant Village data set. These characteristics are categorized using a Bayesian optimum SVM classifier, and the outcomes are evaluated in the form of precision, sensitivity, accuracy, and f-score. The aforementioned approaches will allow farmers all around the world to act quickly to save their crops against suffering irreversible damage, avoiding both a global economic disaster and their own personal financial crisis. The second part of the approach extracts the texture and color information using the HoG, GLCM, and color moments histogram after processing the pictures from the data set. Color, texture, and deep features are the characteristics mixed in this case creating hybrid features Optimization of the binary particle swarm is used to include those hybrid features, and then the simulation outcomes are categorized utilizing a random forest classifier. In hybrid feature selection, binary particle swarm optimization is essential; this algorithm's object is to produce the desired output with the fewest features possible. Utilizing the aforementioned evaluation criteria, a comparison of the two methodologies is performed.