Face detection and facial landmark localization has been an research hotspot in recent years. Face detection is trying to find the general area of faces, while landmark localization is focus on locating the accurate position of facial features, after that we can use these information into other methods such as face recognition. With the rapid development of digital technologies, the task of rapidly finding accurate information of faces from huge digital resources become more and more urgent, that is just what I care about.In this article,the effective and rapid technology of face detection was firstly discussed, Based on the algorithm of face detection of adaboost and LBP,we built a multi-view face detection system consisted of three classifiers, skin color information was also used to filter out non-face region and thus reduce false alarms. Then comes to the facial landmark localization methods, AAM was introduced in this part since its significant performance in this field, but considering speed and accuracy, we choose the CLM algorithm. We also use random forests to initialize the input of CLM considering CLM’s heavily dependency on initialization, thus make these two methods work well. |