In this work, we address a method that is able to track simultaneously head and facial actions like lip and eyebrow movements in a video sequence. In a basic framework, an adaptive appearance model is estimated online by the knowledge of the monocular video sequence. This method uses a 3D model of the face and a facial adaptive texture model. In order to increase the robustness of the tracking, we consider and compare two improved models. First, we use robust statistics in order to downweight the hidden region or outlier pixels. In a second approach, a mixture model provides better integration of occlusions. Each way will be tested separately within the basic framework.