For solving computer vision problems, we are often confronted with high dimensional datasets. The geometric information from manifold learning, which aims to uncover the manifold structure inside the datasets, is useful for solving the application problems. In this presentation, I will discuss some techniques to explore the geometric information, which can be used for solving corresponding computer vision problems. The applications include designing hashing functions for the approximate nearest neighbour searching tasks, recognising the gestures of different video sequences and mining the local structures for clustering problems.
Last modified: Tuesday, 28-Jul-2015 16:18:54 NZST
This page is maintained by the seminar list administrator.