My task is that: there are some text in some position (left, right, top, bottom center, etc) of an images. The style (include size, orientation, font, etc) of text is arbitrary and the content length is arbitrary too.
I want train a classifier to tell whether there is text on the image? The classifier do not need to tell the location of the text.
Can I do that by using convolutional neural network? Is it feasible in principle?