A set of poses (images) was collected and used as a training set. The training stage of the algorithm used principal component analysis (Hotelling Transform) to output a low-dimensional eigenspace on which each pose was represented by its projections.
A set of poses (images) was collected and used as a training set. The training stage of the algorithm used principal component analysis (Hotelling Transform) to output a low-dimensional eigenspace on which each pose was represented by its projections.