I used source code and tested Fine-tuning the detector, but when runned to
canvas = keras_ocr.tools.drawBoxes(image=image, boxes=lines, boxes_format='line')
and
cv2.polylines(img=canvas,
pts=box[np.newaxis].astype('int32'),
color=color,
thickness=thickness,
isClosed=True)
then report an eror like :TypeError: list indices must be integers or slices, not NoneType
I used source code and tested Fine-tuning the detector, but when runned to canvas = keras_ocr.tools.drawBoxes(image=image, boxes=lines, boxes_format='line') and cv2.polylines(img=canvas, pts=box[np.newaxis].astype('int32'), color=color, thickness=thickness, isClosed=True) then report an eror like :TypeError: list indices must be integers or slices, not NoneType