How to fit image (multidimensional array) data into a random forest classifier in python?

Question

I would like to build an image classifier using sklearn.ensemble

I have a list of image X_train where

X_train[0].shape
Out[58]: (353, 1054, 3)

and a list of scalar labels y_train.

Each image X_train[i] is of different shape. When I try to fit these data into the classifier, I get the following error

clf=RandomForestClassifier(n_estimators=100)

clf.fit(X_train,y_train)

ValueError: setting an array element with a sequence

How can I solve this problem?

Answer 1

The most basic solution would be to flatten the image into a single row with an element for each pixel. Do it for the size of the largest image and pad for the others.

This is a bad solution though and will probably not work well.

Image classification requires more sophisticated approaches beyond the scope of this question.