Traditional medicinal plants are types of plants that contain active substances that function to treat and are used by the community to cure or prevent various diseases. Therefore, a study was conducted to test the Local Binary Pattern method for feature extraction of each existing traditional medicinal plant and K-Nearest Neighbor for the classification process after extraction from the Local Binary Pattern method. From testing the Local Binary Pattern and K-Nearest Neighbor methods were able to produce a good accuracy of 96.67%, the accuracy value was obtained from manual convusion matrix calculations with a value of k = 9. Meanwhile, the lowest accuracy results are at k = 1 with an accuracy value of 70%. The extraction and classification results from the Local Binary Pattern and K-Nearest Neighbor methods used 120 datasets which were divided into 90 training data with 6 types of medicinal plant leaves consisting of 15 thorn spinach leaves, 15 binahong leaves, 15 castor leaves, 15 African leaves, and 15 betel leaves with 30 experimental data testing.