This document presents a semi-automatic system for detecting and classifying leaf diseases in soybean plants. The system uses image processing and machine learning techniques. It first segments leaf images into clusters using k-means clustering. It then extracts color and texture features from the clusters. Support vector machines are used to classify leaves as healthy or diseased, and to further classify diseased leaves into categories like downy mildew or leaf blight. The system achieves acceptable average accuracy levels that are better than existing methods. It provides a way to identify leaf diseases early in an automated manner to improve crop yields and food security.