Stress is a part of everyday life and impacts a persons health and well-being in less favorable emotional states such as anxiety, fear or anger. Chronic and left untreated stress can lead to incurable diseases, relationship deterioration and high economic costs. Under the term stress ”the non-specific response of the body to any demand for change”, defined by Hans Selye, is understood. This non-specific response makes it difficult to quantify stress conditions. Stress research developed distinct computational techniques to recognize stress for benefits in a wide range of fields. Mainly physiological parameters are used to diagnose stress. Gathering this health data with the help of contact sensors can be impractically, especially when monitoring a person continuously. Recent studies try to solve this problem based on non-invasive visual computing techniques. The detection of stress based on facial expression models offers promising results. This thesis introduces a solution to detect stress based on a subjects facial expressions. The developed system determines human stress based on extracted facial features from video data. Stress associated facial cues from head, eye, mouth as well as the heart rate from facial photoplethysmography are used to predict stress by machine learning algorithms. A performance evaluation of the developed solution with comparison of existing approaches in literature as well as on annotated data is conducted. Moreover, requirements on facial image data as well as limitations of the implemented system are discussed.