Performance analysis of smile detection using Adaboost algorithms

Face Pose is described as the combination of fixed and movable motions. The fixed and movable motions are associated in such a way that they cannot be separated from one another. Rigid motion is generally related with facial poses, whereas movable motion gets associated with facial expression. For separating one another, a normalized SVD decomposition technique had been utilized. Smile forms a part of facial expression and face pose. Smile detection has many applications in estimating gender, age, etc. This research article discusses two research articles published during 2012 and 2015. These two research articles were proposed novel technique on smile detection using Adaboost algorithms and achieved better performance. They used a database known as GENKI4K for experiment and analysis. Adaboost provided stepwise learning of a nonlinear classification function. It was focused to consider a collection of distinguishing coded features for face recognition. They chosen and combined weak classifiers depending on intensity differences to form a strong classifier. The 2012 research article achieved 89.55% and 2015 research article achieved 96.41% accuracy.