The invention relates to the technical field of computers, in particular to an OCR-based post-loan
inspection method. The method comprises the following steps: S1, shooting and extracting a text picture of a photocopy of customer data; S2, inputting the text picture into a pre-trained text correction
network model; S3, correcting the text picture and the classification
saliency map by utilizing astrip-shaped region
transformation algorithm; S4, outputting an identification text by adopting an optical character
identification technology; S5, determining an error correction word corresponding to an error segmentation word with an
identification error in the identification text; S6, determining an error correction confidence degree corresponding to the error correction candidate text througha neural
network model, and taking the error correction candidate text of which the error correction confidence degree is greater than a first threshold value as an error
correction text; S7, performing information comparison according to the error
correction text, judging whether the information comparison is consistent or not, and if not, prompting manual recheck. According to the invention, the technical problem that the content of a photoprint with poor photoprint quality is inaccurately identified and captured by using an OCR technology is solved.