Bbox,confidence=cv.text_TextDetector.detect(inputImage)

edit

cv::text::TextDetector is an abstract interface. you have to create an instance of the actual implementation:

textSpotter = cv.text.TextDetectorCNN_create("textbox.prototxt", "TextBoxes_icdar13.caffemodel")

then you can detect text segments with it:

rects, outProbs = textSpotter.detect(img);

there is a full python example here , have a look !

data needed: caffemodel prototxt

also make sure, your cv2 was built with the opencv_contrib modules !