TypeError: only length-1 arrays can be converted to Python scalars about opencv-python-blueprints

[mbeyeler]

Thanks for posting. It is not immediately clear to me what is causing this error. It seems to be caused by the train function of the OpenCV classifier.

I suspect either X_train or y_train_bin on L256 has length 1, and that is causing some issues with the classifier's train function. Can you add a print statement right before that to see what X_train[data_id, :], y_train_bin, and self.params look like?

[Algabri]

Thanks, @mbeyeler
The outputs are:

X_train [[  85.   96.  127. ...,  100.  104.  113.]
 [  19.   18.   17. ...,   71.  103.  132.]
 [  93.   92.  186. ...,   11.    8.    9.]
 ..., 
 [  22.   20.   21. ...,   76.   62.   44.]
 [  30.   29.   30. ...,   28.   28.   29.]
 [  75.   89.   90. ...,  241.  225.  233.]]
 - strategy one-vs-one
    - train
y_train_bin [0 1 0 ..., 1 1 0]

[mbeyeler]

Looks all good. I think the problem is that they changed the OpenCV API to train(samples, layout, responses), similar to what is explained here.

Try adding a layout parameter to L256:

self.classifiers[svm_id].train(X_train[data_id, :], cv2.ml.ROW_SAMPLE, y_train_bin)

[Algabri]

Thanks, yes.
It is fine now.
Sorry, @mbeyeler. I wrote this answer before appearing this error.
I already change it to self.classifiers.append(cv2.ml.SVM_create()) but it is still giving error.
Another error:

Traceback (most recent call last):
  File "/home/redhwan/Downloads/opencv-python-blueprints-master/chapter6/chapter6.py", line 78, in <module>
    main()
  File "/home/redhwan/Downloads/opencv-python-blueprints-master/chapter6/chapter6.py", line 49, in main
    acc, prec, rec = MCS.evaluate(X_test, y_test)
  File "/home/redhwan/Downloads/opencv-python-blueprints-master/chapter6/classifiers.py", line 298, in evaluate
    y_hat = self.classifiers[svm_id].predict_all(X_test_id)
AttributeError: 'cv2.ml_SVM' object has no attribute 'predict_all'