Audience engagement is a key endpoint of the entire #moviemaking process, so moviemakers are in constant need of finding better ways to #gaugeaudiencereactions. Did they enjoy it? If they did then; how much? Scientists from Caltech and Disney Research have developed a system that uses a #facialexpressiontrackingneuralnetwork to study and predict how individual members of the audience react. This method allows for the real time tracking of facial expressions in a reliable and relatively simple manner.
The researchers collected large datasets of #facialexpressiondata by recording hundreds of people watching movies using an #infraredhidefcamera which captures everyone’s expressions and motions. The resulting data was fed to a neural network. Once the neural network had finished training, the team validated model using real-time audience footage in and attempted to predict the expression a particular individual would make at various time points. They researchers found that this neural network could reliably #predictlaughsandsmiles of the audience.
Applications of such technologies can be employed in other scenarios like #crowdmonitoring, or interpreting complex visual data in real-time.Tags: Artificial Intelligence, artificial neural networks, crowd monitoring, Face Recognition, facial expression tracking, machine learning