Timezone: »

Deep Structured Prediction for Facial Landmark Detection
Lisha Chen · Hui Su · Qiang Ji

Tue Dec 10 05:30 PM -- 07:30 PM (PST) @ East Exhibition Hall B + C #65

Existing deep learning based facial landmark detection methods have achieved excellent performance. These methods, however, do not explicitly embed the structural dependencies among landmark points. They hence cannot preserve the geometric relationships between landmark points or generalize well to challenging conditions or unseen data. This paper proposes a method for deep structured facial landmark detection based on combining a deep Convolutional Network with a Conditional Random Field. We demonstrate its superior performance to existing state-of-the-art techniques in facial landmark detection, especially a better generalization ability on challenging datasets that include large pose and occlusion.

Author Information

Lisha Chen (Rensselaer Polytechnic Institute)
Hui Su (IBM and RPI)

Hui Su is the Director of Cognitive and Immersive Systems Lab (CISL), a research collaboration between IBM and Rensselaer Polytechnic Institute (RPI). He is now a Professor of Practice in Computer Science Department in RPI, and has been an executive and technical leader in IBM Research. During 2012-2015, he was the director of IBM Research Lab in Cambridge, MA., responsible for a broad scope of global missions, including Cognitive User Experience, Center for Innovation in Visual Analytics and Center for Social Business. He was also once the Associate Director of IBM China Research Lab in 2011-2012, leading a research team with more than 100 researchers working on speech recognition, natural language understanding, human computer interaction, healthcare research and cloud computing. As a technical leader and a researcher for 23 years in IBM Research, Hui Su has been an expert in multiple areas ranging from Human Computer Interaction, Cloud Computing, Visual Analytics, Neural Network Algorithms for Image Recognition etc. Hui has dozens of patents and technologies developed in these areas.

Qiang Ji (Rensselaer Polytechnic Institute)

More from the Same Authors