futurely
Results
22
comments of
futurely
The individually learned multi-view models are even worse.  
The test images are the IBUG database downloaded from [here](http://ibug.doc.ic.ac.uk/resources/facial-point-annotations/).