Object Class Invariant model (OCI)

OCI: seems to be a common way to deal with the multi-view object recognition task, especially used by researchers working in the area of face detection / identification


The viewpoint invariant approach relates features in different images via an object class invariant (OCI), a geometrical reference frame that is uniquely defined for each object class instance and invariant to projective image transform arising from viewpoint change. As the variable of viewpoint is effectively marginalized from the formulation, a viewpoint invariant model can be learned from natural imagery taken from arbitrary viewpoints from labeled images.

 
public/object_class_invariant_model_oci.txt · Last modified: 2012/09/05 15:01 (external edit) · []
Recent changes RSS feed Powered by PHP Valid XHTML 1.0 Valid CSS Driven by DokuWiki