Ask Your Question

Revision history [back]

I have the same problem. Open CV 3.4.1, C++. Feeding this csv to train the model:

./faces/face003.jpg 3
./faces/face005.jpg 4
./faces/face009.jpg 7
./faces/face011.jpg 8
./faces/face014.jpg 10
./faces/face016.jpg 11
./faces/face018.jpg 12
./faces/face019.jpg 25
./faces/face021.jpg 13

All 3 face recognizers (LBPHF, Eigen and Fisher) train OK and output a prediction when fed with a test image.

If I change the csv to for example:

./faces/face019.jpg 12

(Label 12 now has 2 reference images face018.jpg and face019.jpg) Then I get the same error message as the OP.

The other recognizers do not throw that exception when trained with the same data set which kind of hints that the problem does not seem to be with the training set but with the training algorithm used by FisherFace.