_______________________________________________________________________________

3D reconstruction and modeling of urban buildings and architecture

Objective
To realize automatic/semi-automatic reconstruction and modeling of urban building or architecture with complex components

Challenges

1. Photo-realistic & geometrically accurate
2. Multi-images of target from different vantage points may not always be available

Approach

Employ constraints derived from shape symmetry

1. Camera calibration by exploiting symmetry
2. Duplication of virtual cameras
3. Reconstruction by combining real and virtual pairs
4. Modeling according to shape symmetry

_______________________________________________________________________________

Vision-based human activity analysis

Objective

To recognize common (single or dual actor) actions in natural video sequences

Challenges

1. To extract motion information in video
2. To model temporal motions and their relationships

Approach

1. Motion information encoded in motion trajectories
2. Hierarchical levels of info extracted: appearance, intra-trajectories, and inter-trajectories

_______________________________________________________________________________

Contextualized histograms for group activity analysis

Objective

To analyze group activities that involve multiple actors with interactions

Challenges

1. To model interaction between different actors
2. To keep the model complexity low

Approaches

1. Raw motion information encoded in higher order local statistics
2. Randomized forest employed to facilitate model simplification with supervision