• FEATURED
  • PUBLICATIONS
  • SOFTWARES
  • ABOUT

Factorized Graph Matching is proposed for better optimizing and understanding graph matching problems.

Graph Matching

Canonical Time Warping is proposed for temporally aligning multi-modal sequences.

Temporal Alignment

Aligned Cluster Analysis is proposed to temporally cluster human behavior.

Temporal Clustering

An Unsupervised System is proposed to discover facial events from video of naturally occurring facial behavior.

Facial Event Discovery

Exemplar-based Graph Matching is proposed to robustly localize facial landmarks on image.

Facial Landmark Localization

Time Mapping generates low-frame-rate video from a high-frame-rate input using Video Saliency.

Time Mapping & Video Saliency

Spatio-temporal Matching is proposed for detecting and tracking humans in videos.

Human Pose Estimation

Bipartite-Graph Labels is proposed to exploit the rich relationships among fine-grained object classes.

Fine-graind Object Recognition

MagFace learns a universal feature embedding whose magnitude can measure the quality of the given face.

Face Recognition