Spotting all the text in the image with line-level, and output in JSON format.