FaceReader works in three steps

FaceReader™ is the most robust automated system for the recognition of a number of specific properties in facial images, including the expressions:

  • happy
  • sad
  • angry
  • surprised
  • scared
  • disgusted
  • 'neutral' state

 
FaceReader works in three steps: 1) face finding, 2) face modeling, and 3) face classification. Initially, an accurate position of the face is found using the Active Template Method. Next, the Active Appearance Model is used to synthesize an artificial face model, which describes the location of 491 key points as well as the texture of the face. As a result FaceReader can accurately analyze facial expressions as the test participant moves their head. Finally, the face classification delivers the output of six basic expressions and one neutral state.


Automatic calibration
When desired, you can use the automatic calibration to tailor the analysis of facial expressions to a specific person. You can either run the calibration before the analysis or continuously during the observation.


Additional classifiers
In addition to the 6 basic expressions, FaceReader also automatically classifies mouth open-closed, eyes open-shut, and eyebrows raised-neutral-lowered. This feature will save you valuable time when coding video. FaceReader also registers head orientation and gaze direction. This provides you with valuable supplementary data in addition to the facial expression analysis.


Video stream, image or live analysis
You can work off-line using video, on-line for live analysis using a USB webcam or IP camera, or you can upload still images for analysis. If you plan to analyze multiple videos, you can analyze them in a group. Once you select the videos to be analyzed, the software does the work for you. 
 
When analyzing from video, you can choose an accurate frame-by-frame mode or skip frames for high-speed analysis. The best results are achieved using diffused frontal lighting on the test participant; Noldus provides illumination for optimization of your set-up. You can follow the emotions of your test participants if their orientation, movement, and rotation is within certain limits.

FaceReader is a user friendly package which helps you automate your research. The software can quickly detect interesting episodes, even in long series of events. The computer does the work for you! You can choose to analyze the whole video or only parts of it. FaceReader includes a basic person identification algorithm, allowing subjects to be recognized after initial input of the original facial image. After the analysis, you can export data together with the identity profiles.


Visualization
FaceReader automatically analyzes facial expressions, which can be visualized as bar graphs, in a pie chart, and as a continuous signal. A gauge display summarizes the negativity or positivity of the emotion (valence).The timeline gives you a detailed visual representation of the data. A separate reporting window displays a pie chart with percentages, a smiley, and a traffic light, indicating whether a person’s mood is positive, neutral, or negative. All visualizations are given to you in real-time and may be viewed afterwards. 

In addition to analyzing expressions, FaceReader can classify faces based on the following characteristics: gender, age, ethnicity, and facial hair (beard and/or moustache). FaceReader can accurately read the faces of children who are 3 years or older and elderly people. Although you can analyze all ethnic groups, FaceReader is currently not designed to work with children from East Asia or South-East Asia. The test database is still being expanded. Glasses may also interfere with the analysis process.