Machine Learning for OpenCV 4 - Second Edition

Constructing the tree

Building the decision tree with OpenCV works in much the same way as in Chapter 3, First Steps in Supervised Learning. Recall that all of the machine learning functions reside in OpenCV 3.1's ml module:

1. We can create an empty decision tree using the following code:

In [16]: import cv2...      dtree = cv2.ml.dtree_create()

2. To train the decision tree on the training data, we use the train method. That's why we converted the data to float before—so that we could use it in the train method:

In [17]: dtree.train(X_train, cv2.ml.ROW_SAMPLE, y_train)

Here, we have to specify whether the data samples in X_train occupy the rows (using cv2.ml.ROW_SAMPLE) or the columns (cv2.ml.COL_SAMPLE).

3. Then, we can predict the labels of the ...