OpenCV 4 with Python Blueprints - Second Edition

Testing the MLP

Analogous to Chapter 7, Learning to Recognize Traffic Signs, we will evaluate the performance of our classifier in terms of accuracy, precision, and recall.

To reuse our previous code, we just need to calculate y_hat and pass y_true alongside it to the metric functions by doing the following:

1. First, we featurize our test data using the pca_args we stored when we featurized the training data, and the _pca_featurize function, like this:

     x_test = _pca_featurize(np.array(data)[test], *pca_args)

2. Then, we predict the new labels, like this:

_, predicted = mlp.predict(x_test)    y_hat = np.array([index_to_label[np.argmax(y)] for y     in predicte

3. Finally, we extract the true test labels using indices we stored for testing, as follows: ...