Facial Action Unit Recognition using Temporal Templates and Particle Filtering with Factorized Likelihoods

Automatic recognition of human facial expressions is a challenging problem with many applications in human-computer interaction. Most of the existing facial expression analyzers succeed only in recognizing a few basic emotions, such as anger or happiness. In contrast, the system we wish to demonstrate recognizes a large range of facial behavior by recognizing facial action units (AUs, i.e. atomic facial signals). Our system performs AU recognition using temporal templates as input data and a combined kNN-rulebase two-stage classifier. Besides demonstrating the facial action recognizer, we will demonstrate a new point-tracking algorithm based on particle filtering with factorized likelihoods that we use for the registration of the input data.