Glossary
Optical character recognition (OCR) converts text visible in an image or scanned document into machine-readable text. It helps make product labels, scanned catalogues and image-based datasheets searchable or available for extraction. OCR recognises characters and layout; a further step is usually needed to identify the product, interpret attributes and validate their values.
A PDF can contain selectable digital text, scanned images or a mixture of both. OCR is useful where the needed text is only present as pixels. Processing can also return locations, page structure and confidence information that help an application relate extracted text to the original page.
For example, OCR may recognise a dimension table in a scanned motor datasheet. The next stage still needs to associate each value with the correct row, column, unit and product variant. Reading the characters correctly does not by itself resolve those relationships.
Low resolution, skew, small type and complex tables can create errors. Characters such as zero and the letter O may be confused, while a unit or minus sign can be lost. Retain the original file and relevant page location so a reviewer can inspect the result.
Check product matching, character recognition and semantic extraction separately. Use available digital text where appropriate and assess the actual document quality. A high OCR confidence score concerns the recognition task; it should not be treated as proof that the proposed product attribute is true.
Discuss the product information and controls your AI use case needs.