Paddy is a crucial agroculture sector since rice is the staple food for the majority of the world's population. However, the production of paddy is slower and less productive since many factors have affected the growth of the paddy. The existence of disease in paddy component affects the quality of rice produced. Hence, the recognition of the disease at the beginning stage is crucial as the initial approach for prevention purposes. In this study, a system is developed to detect the paddy leaf disease such as bacterial leaf blight, brown spot and leaf smut. All the processes involved are implemented and compiled using MATLAB R2020a. A set of 105 image data with disease is converted to binary image using thresholding. 6 features from all the data are extracted and divided to testing and training set before the classification process. A cubic support vector machine is used for the classification process. Lastly, accuracy, precision, and misclassification for each disease are calculated for performance evaluation. Results show that the average performance of the diseases on accuracy, precision, and misclassification are 88.57%, 82.97%, and 11.43% respectively. The use of the processes act as assistance to the paddy farmer to identify the existence of the paddy leaf disease. This could improve the quality of the paddy produced by reducing the process of manual disease checking.