I run opencv_traincascade for say 20 stages. It fails at stage 15 saying:
Required leaf false alarm rate achieved. Branch training terminated.
Than I run opencv_traincascade for 15 stages, this time it fails at stage 13. Then I run for 13 stages, this time it fails at stage 12 etc..
How can this happen? Looking at output values are same at each run, but fails at different stages.