When copying a product code or reference from an image, readable output is not a verified identifier. Before searching inventory, completing a document or forwarding it, compare the result with the original character by character.
Check codes separately from sentences
Tesseract documents its default recognition as optimized for sentences of words; codes are a different kind of input. FileMyLove extracts text but does not validate a product identifier against a catalog or check-digit rule.
These are suggested proofreading checks, not measured error rates for the tool.
| Check | Suggested action |
|---|---|
O versus 0 |
Inspect that position in the source |
I, l versus 1 |
Do not decide from a word's apparent meaning |
| Leading zeros | Compare every character from the beginning |
| Hyphens and spaces | Follow the source owner's format rather than deleting all separators |
Start with a readable image
Tesseract identifies noise and skew as recognition problems. Use the clearest original, straighten the text, and crop the relevant region with a small margin rather than cutting into characters.
If you cannot distinguish the character yourself, request a clearer image or source text. Repeated OCR results are not independent confirmation of accuracy.
Proofread before copying
- Open Image to text and choose the appropriate language.
- Keep the original image beside the extracted text.
- Compare each position, including the first and last characters and separators.
- Where available, verify against a trusted source list.
- Correct only confirmed differences and set uncertain entries aside.
Avoid a global replacement of O with 0: it can damage correct entries. The current interface offers neither a custom character whitelist nor automatic code validation.
Separate later changes from recognition errors
If a correct copied identifier changes after spreadsheet import, investigate that step separately. See long identifiers in Excel and preparing Thai screenshot text for OCR.
