Smartling uses OCR (Optical Character Recognition) to automatically match the text in image context files to your content. You can choose to match the text in the image with strings in your entire project, or in a specific file in the project, or even disable OCR for an upload.
For OCR to work, make sure your source content (strings) have been uploaded to Smartling before you upload your screenshot. Continue reading for steps on how to troubleshoot OCR.
As soon as an image file is uploaded to Smartling, and matching preferences are decided for the upload, Smartling performs OCR on the image file to automatically match the text in image context files to your content. When there is a successful match, Smartling creates a binding between the text in the image context and the matched string in the content. The result of this is highlighted strings in the Context panel in the CAT Tool.
Occasionally, OCR may not map strings automatically. There can be a number of reasons why the OCR has not worked as intended. Here are some tips on why OCR may not work;
- The image is of poor quality / small size file
- The image contains unusual fonts
- The image is missing some terms that are included in the string
- The image contains special characters that are not included in the string
- The image contains text that is blurry or small
- The string is too long
Here are some tips to ensure Visual Context is mapped correctly;
- Ensure the image is high quality and focused.
- Ensure the image and string contains the same text, in a standard font, with no special characters
- Always check the matching by clicking the context thumbnail to view the context in detail and remove any mismatched strings
- Manually map any unmapped strings to ensure Visual Context for Translators
Tip: For guidance on how to review OCR and manually modifying matching, read our documentation on Manually Match Strings to Visual Context in the Context Editor.