This is the evaluation code for tesseract and parseq OCR apps. Current performance is very poor, and I am unsure as to whether or not that is a problem with the eval code or external factors (EAST, model performance, etc). The common errors with Parseq (test output that is correct but truncated) imply that it's a bounding box issue, whereas the tesseract output is more chaotic.
This is the evaluation code for
tesseractandparseqOCR apps. Current performance is very poor, and I am unsure as to whether or not that is a problem with the eval code or external factors (EAST, model performance, etc). The common errors with Parseq (test output that is correct but truncated) imply that it's a bounding box issue, whereas the tesseract output is more chaotic.