Shape recognition

This course deals with pattern recognition using classical descriptors ("region" and "contour") and classification methods (Machine Learning).

Descriptors

Shape descriptors:

Freeman/Signature/Fourier contour approaches
Structure/geometric region approaches
Hough transform


Pattern descriptors:

Points of interest: Harris/SIFT
Dense: LBP, ...
Block-wise: HOG, ...


Dimension reduction:

PCA






Classification methods/Machine Learning

Unsupervised classification methods

Hierarchical clustering
K-means clustering methods


Supervised classification methods

Linear/quadratic disriminant analysis
K-nearest neighbors
SVM
Bags of visual words