Group items tagged

This comes out of work I presented as a SIGGRAPH 97 Technical Sketch. The demo is mostly an experiment to isolate the minimal number of facial expression elements that will produce a "convincing" impression of character and personality. Of course this is just a subset of the full range of expression the human face is capable of. Paul Ekman's pioneering work on the Facial Action Coding System gives a functional description of the full range of expression of which the human face is capable.

I did send it a year a go, and have been looking at it off and on since it was created. As a simple base set of points it's wonderfully expressive and shows how much reads with so little adjusted. Head angle and noise plays a big part in this too.

I was playing with that. Very cool! It just shows the system, simplied as possible. What is "heas" angle?

FaceTracker is a C/C++ API for real time generic non-rigid face alignment and tracking. Goal: Non-rigid face alignment and tracking is a common problem in computer vision. It is the front-end to many algorithms that require registration, for example face and expression recognition. However, those working on algorithms for these higher level tasks are often unfamiliar with the tools and peculiarities regarding non-rigid registration (i.e. pure machine learning scientists, psychologists, etc.). Even those directly working on face alignment and tracking often find implementing an algorithm from published work to be a daunting task, not least because baseline code against which performance claims can be assessed does not exist. As such, the goal of FaceTracker is to provide source code and pre-trained models that can be used out-of-the-box, for the dual purpose of: 1. Promoting the advancement of higher level inference algorithms that require registration. 2. Providing baseline code to promote quantitative improvements in face registration. Features: * Real time: ranging from 20-30 fps (depending on processor, compiler and use of OpenMP) * Generic: designed to work for most people under most conditions * No training required: a pre-trained model is provided * Detection based initialisation: no user intervention required * Automatic failure detection: requires no user re-initialisation * Camera or video input

Thanks Maurizio for the link!