Human faces contain rich information. Recent studies found that facial features have relation with human weight or body mass index (BMI). Decoding “facial information” from the face in predicting the BMI could be linked to the various health marker. This paper proposed the classification of body mass index (BMI) using facial landmark approach based on facial images. In this framework, Discriminative Response Map Fitting (DRMF) method has been used as feature extraction technique to detect and locate the facial landmark points on the facial images. About sixty-six (66) facial landmark points were identified. Only nineteen (19) of facial landmark points have been employed to extract the facial features in terms of distance and ratio features. A total of 221 facial landmark features were obtained and used as feature vector to classify the BMI classes. The rationale of using 221 facial landmark features is because these features were able to exhibit the unique characteristic of the BMI classes, which are normal, overweight and obese. Then, the extracted features were further reduced using Principal Component Analysis (PCA) plus Linear Discriminant Analysis (LDA) to map high dimension features into low dimensional feature with maximize between class scatter and minimize within class variations. Later, the reduced features were subjected to k-NN classifiers. A series of experiments has been conducted on MORPH II database using the reduced facial landmark features to classify the three BMI classes. Based on the experimental results, it shows that the reduced features using PCA plus LDA based on k-NN classifier has achieve the highest recognition rate with accuracy of 83.33 %. The obtained results show that the reduced facial landmark features were able to discriminate the three BMI classes of normal, overweight and obese, thus shows the promising results.
