Facial Expression Detection Using Local Binary Pattern And K-Nearest Neighbor Methods

Abstract

Facial Expression Detection is the recognition of a pattern where the input is a digital image and the output is a label of a person's emotions that have been made into a class, which class has been stored in the database as training data to find the closest or similar. Pattern recognition with training data or similar classes is done using artificial intelligence with various methods. This study aims to test the Local Binary Pattern and k-Nearest Neighbors methods to be implemented in facial expression detection and create a system on a computer to be able to know human facial expressions are happy or sad. Local Binary Pattern is defined as the ratio of the binary value of the pixel at the center of the image to the 8 values of the surrounding pixels. K-Nearest Neighbors algorithm with supervised learning which aims to find new patterns in the data by connecting new data patterns with existing data patterns. Based on the results of manual testing on sad expressions, the accuracy is 90% and happy expressions are 80%. Furthermore, the K- Fold Cross Validation test, at 5-Fold Cross Validation at 61.66% and at 10-Fold Cross Validation at 75%.