This forum is disabled, please visit https://forum.opencv.org
 | 1 | initial version |
you can see here, that only C-1 components are retained in the LDA (where C is the number of classes). if you have 3 persons, you can only have a maximum of 2 projections.
maybe it gets easier to understand, if you think of the LDA row vecs as "border" or "difference" representation between classes, for 2 classes, you need 1 border, for 3 classes 2 borders, etc.
(the example code was working on att faces with 40 individuals)
| | 2 | No.2 Revision |
you can see here, that only C-1 components are retained in the LDA (where C is the number of classes). if you have 3 persons, you can only have a maximum of 2 projections.
maybe it gets easier to understand, if you think of the LDA row vecs as "border" or "difference" representation between classes, for 2 classes, you need 1 border, for 3 classes 2 borders, etc.
(the example code was working on att faces with 40 individuals)
"but i don't see them updated"
you'll only get different images, if you retrain it with different input data, it is not using the signal from the camera at all.
