Infoscience

Report

Facial Expressions Evaluation Survey

Facial expressions are probably the most visual method to convey emotions and one of the most powerful means to relate to each other. A typical automatic system for the recognition of facial expressions is based on a representation of the expression, learned from a training set of pre-selected meaningful features. The learning process relies on the labels associated by an expert or a group of experts to the training samples. The experts are asked to associate each images in the training set to one of the expressions we are dealing with. In other words we must have label makers (the experts) reliable enough and who have strong knowledge of the problem in order to ensure the correctness of what we are trying to learn. What is really important is to how get and use this knowledge. The facial expressions evaluation survey is born in order to find a way to extract this knowledge directly from the experts. In the issue of expressions evaluation every single human can be considered as an expert and gives his/her contribution in building this ”common sense knowledge”.

Related material