Research Assistant Intern 0939
February 17, 2009
Project Name
Graphics and Computer vision: The facial display database
Project Description
Two important visual cues for human interactions are facial expressions and eye gaze patterns. Human participants use these facial displays to express emotion (e.g., confusion, surprise or acknowledgment), turn taking (e.g., look back at the other participant) and state of mind (e.g., eye gaze aversion expressing thinking). To be able to recognize these facial displays, we need an exhaustive database which contains enough samples/examples to represent every variation found in real life scenario: individuals (variation in appearance between people), lighting (change of lighting conditions), and pose (seen from different view point). The goal of this project is to build a new video and image database of human facial displays using the pre-existing Light Stage developed at the ICT. Using high speed cameras synchronized with a light sphere, the Light Stage can record the same facial display under varying illumination and pose.
Job Description
The main goal of this internship is to assist with the design, acquisition and post-analysis of the facial display database. At the design stage, the intern will evaluate different approaches for evoking facial expressions and emotions from the participants. It will be important to create a protocol describing the procedure for data acquisition. During the acquisition stage, the intern will be in charge to schedule the participants and assist with the acquisition process (under the supervision of Graphics lab members). Finally the post analysis stage will open the door to some preliminary experiments using this data: building a multi-view appearance model of human facial display.
Skills required/preferred:
• Familiar with facial expression analysis and gaze estimation
• Familiar with computer vision and appearance model
• Programming knowledge in C++ and Matlab
• Experience with codec and video encoding/decoding
• Strong verbal and written communication skills