Face Alignment by Heng Yang

Date: Wednesday, March 12th 2014
Time: 4-5pm
Place: Informatics Teaching Lab (ITL) — Top floor seminar room

Speaker: Heng Yang

Title: Regression Forests for Face Alignment

Face alignment, in other words, the localization of a set of facial landmarks in an image, is a well studied topic in the computer vision literature. It is often the first step for face recognition, facial animation and facial expression recognition. However, the current methods still have difficulties in dealing with face images in the wild, with low quality, head pose variation and partial occlusions, especially when real-time detection is needed.

In this talk he will briefly describe the general framework of regression forests, that has emerged as a very effective method in various computer vision tasks (the most famous application is the human pose esitmation in Kinect). This talk will focus on its application on face alignment, based on his previous work (H. Yang and I. Patras, Sieving Regression Forest Votes for Facial Feature Detection in the Wild, ICCV2013).

Short Bio:
Heng Yang is a 3rd year PhD student in Multimedia and Vision Group, working with Dr. Ioannis Patras. His research focuses on random forests, cascaded pose regression and face analysis. http://www.eecs.qmul.ac.uk/~hy300/​