Wrong results using the neural network classifier
Models trained with TrainVectorClassifier
using the ann classifier (OpenCV neural network) produce wrong results. When the application computes the confusion matrix after the training part, all samples are classified as the first class, even when using the training dataset as validation dataset.