I'm using openALPR library to read plate licenses but I'm having issues reading the plates in different angles, like below image. My question is: what's the proper way to do that? Image processing to try make the image straight before submitting such as Homography ?openALPR's settings such as max_plate_angle_degrees max_detection_input_width, max_detection_input_height, etc or train the tesseract-ocr with cropped images in different angles? I have no code to show because I'm looking for a direction how to do that.
