April 2020
Intermediate to advanced
438 pages
12h 2m
English
Running a forward pass on the pretrained EAST model returns scores and geometry that are decoded using the decode_predictions() function to obtain the bounding boxes (ROIs) predicted to be containing text. Next, the texts inside these ROIs are to be extracted with the pytesseract image_to_string() method.
To extract texts as strings using Tesseract v4 OCR, the command-line arguments are needed to be passed as a configuration to the image_to_string() method of pytesseract (for example, we used "-l eng --oem 1 --psm 11" as the configuration):
Read now
Unlock full access