After the scanned document is translated, the translation is empty or missing paragraphs. How to troubleshoot?

Key points of this article

If the translation is empty, first check whether the original text can be selected. If not, it means that the image is scanned and text recognition needs to be completed first.
Missing paragraphs are mostly caused by blurred, tilted, shadowed or sealed pages. Please edit the image first and then resubmit it.
When you only see part of the content, check whether only part of the page has been submitted, and use bilingual comparison to check the left and right sides
When the translation is incomplete, separate the problematic page and resubmit it. It is easier to locate the cause than re-doing the entire translation.
After translating important content, you need to compare it with the original text paragraph by paragraph. Both machine translation and text recognition may make mistakes.

The translation of the scanned document is empty, or the content is missing halfway through the page. Many people subconsciously think "this tool is not good". In fact, in most cases, the problem lies in a certain link not being completed properly. Scans are essentially pictures, and the text in the pictures needs to be read out before translation. If this step fails, is only partially successful, or the page itself is not of sufficient quality, it will eventually appear as blank or missing paragraphs.

This article is written for readers who are dealing with scanning contracts, scanning papers, and archiving paper materials. After reading, you should be able to determine whether the problem lies in the file, text recognition or output settings, and complete the troubleshooting and resubmit in order. It is a part of the troubleshooting method for document translation failure that specifically targets scanned documents. It is suitable for online translation and PDF tools in desktop software.

If you are searching for "free online scanned document translation" or "which scanned document translation software is the best", it is recommended that you complete the following troubleshooting before deciding whether to change tools. Changing the same file to another tool often results in the same result, because most of the problems lie in the files and settings.

The translation of the scanned document is empty or missing paragraphs, how to troubleshoot? One sentence conclusion

In most cases, first confirm whether the text can be selected in the original text, whether the text recognition is completed, and whether the page is clear, and then look at the output options. Do not resubmit the entire document until the problem is clearly identified.

your situation	Recommended practices	reason
The entire translation is blank	First check whether the original text can be selected, and then confirm whether the file is complete and can be opened normally.	If the entire file is empty, it usually means that the file has not been read correctly, or the overall text recognition has not been completed.
Just a few pages missing	Check the submitted page range and separate the missing pages and resubmit them.	When individual pages are of poor quality or have special structures, they may easily fall out.
Paragraphs disappear in pieces	Check whether there are shadows, seals, or oblique shots on the original page, correct them and try again.	Blurred or occluded images will cause recognition to skip the corresponding area
The translation does not match the original text	Switch to bilingual comparison, compare left and right paragraph by paragraph	It’s impossible to tell what’s missing just by looking at the translation. Bilingual comparison makes it easier to locate.
The file is large and has many pages	Split into several files according to chapters and submit them separately.	The larger the file, the harder it is to find the problematic paragraphs in a single processing, and it is easier to lock the range after splitting.

First determine what kind of problem it is

Before troubleshooting, first check the table below to find similar symptoms, and then take action after confirming the cause.

symptoms	Common causes	How to confirm
The translation is completely blank	Text recognition is not executed, or the file is encrypted or damaged.	Open the original file with a reader and try to select a section of text. It can be opened but cannot be selected, indicating that it is a picture page.
Only the first few pages have translations	Only some pages were included during submission, or subsequent pages were abnormal.	Check the page range submitted and see if the last few pages of the original text are clear.
Some paragraphs are missing	Pages are blurry, tilted, shadowed, or obstructed by stamps	Zoom in on the original page and see if there are any of the above conditions near the missing paragraph.
Table content is misaligned or blank	Table lines are unclear, or the table is broken across pages	Compare with the original page to check if the table borders are complete and if the edges are broken
The translation looks like it hasn't been translated	Only the translation output is selected, and the original recognition result itself is empty	Switch to bilingual comparison and reread to see if the original text on the left side has content
The formula and diagram captions are missing	Formulas and image areas are skipped as images	Confirm these locations by comparing them with the original page, and check these points separately

Inspect and fix each item one by one

Follow the steps below; the earlier steps are simpler and can resolve most issues.

Confirm that there are no issues with the documents themselves. Open the original file with a reader and check whether all pages display properly and whether there is password protection. If the file was downloaded from the internet or email, it may be cut off during transmission, and you can obtain a new copy of the original file.
Test whether the original text is an image. Try selecting a passage in the reader and copying it. If you can't select it, it means it's a scanned image, and you must first complete text recognition.
Confirm that text recognition has been completed. After uploading the scanned PDF in omnidoc.top, wait until processing is complete before downloading the result to avoid downloading incomplete files during processing. If you are using a desktop client, you can first use the OCR function in the PDF editor to process the file, then check if the text is selected, and only submit the translation after confirming.
Check page scope and output options. Make sure no pages are missed during submission, then select bilingual output first. In bilingual results, the left side shows the original text recognition content; if the left side is missing, the problem lies in the recognition stage; If the left side is complete and the right side is missing, the problem is more likely to lie in the translation process.
Separate the problem page and resubmit it. Use desktop client page operations to split missing or abnormal pages and submit them separately. It's easier to verify the results of small files and quickly confirm whether the issue is with these few pages.
Fix page quality and try again. Rotate and correct slanted pages first; For shadowed and blurry pages, it is recommended to re-scan using a clearer method, ensuring even lighting, flat paper, and not pressing down on binding edges.
Compare segment by paragraph and record the locations of missing segments. Compare the translation and the original text page by page, and note down the page numbers and paragraphs where the missing paragraphs are. If there are still missing pages after reprocessing, you can note the page numbers and paragraph positions to decide whether to re-scan or replace the document.

How to fix poor page quality?

Common quality issues with scans fall into three categories: tilting, shadows and occlusion, and low contrast. Tilt can be rotated and corrected before submission. Shadows come from paper lifting or uneven lighting; when rescanning, spread the file flat and switch to a uniform light source usually yields better results than post-processing. If the seal or handwritten annotations are pressed on the main text, they will cover the text. In this case, either avoid them on the original or separately indicate the seal area before scanning, and pay special attention to these positions during verification.

Output option causes "Looks missed"

Sometimes the content isn't lost, just the wrong way to output it. Only the translation output shows the result. If a certain paragraph does not appear, it is impossible to determine whether the recognition was missed or if the translation step was skipped. Bilingual comparison places the original and translated text side by side, allowing you to directly see which paragraph contains only the original text and no translation. During the troubleshooting phase, it is recommended to first use bilingual reference to confirm the issue before switching to translation-only output.

How to Prevent It from Happening Next Time

When scanning, lay the paper flat and light evenly, avoiding creases and shadows, and try to keep each page facing the camera as closely as possible.
Before submitting, use a reader to check whether the text is selected, to determine whether it is a text version or an image version, and decide whether text recognition is needed first.
Long documents are submitted by chapter or by several pages, with page number records kept to quickly locate issues when they arise.
When processing a certain type of scanned document for the first time, first spot check the beginning, middle, and last few pages to confirm no missing paragraphs, then process the entire document.
Important content such as contract terms, thesis data, and medication descriptions must be translated and checked paragraph by paragraph with the original text. Machine translation and text recognition can both make mistakes, so do not use them directly as the final version.
Keep the original scanned copy; don't just save the translation. When problems arise, you can compare them with the original documents to determine which step went wrong.

❓ Frequently Asked Questions (FAQ)

1. What should you look for when choosing a free online tool for scanned document translation?

First, check the file size and page limits, then check if the translation has watermarks and how to handle the uploaded files. Whether formatting, tables, and bilingual reference are supported also affects whether they can be used directly. It is recommended to test with a few pages of samples first, and only process important documents after confirming the results are usable.

2. Which scanned document translation software is the best to use?

Whether it's good to use depends on the type of document you have and your needs. For scanned PDFs, the key is whether it can recognize text first, then retain the layout for output. You can use several pages of the same sample to separately check the blanks, missing paragraphs, and layout; the results can illustrate the problem.

3. After re-translating, the paragraph still misses—is this a tool problem?

Not necessarily. First, make sure the original page is clear, text recognition is complete, and the page range is not missed. If all three items are normal but the missing segments persist, consider changing tools or replacing the scan. Record the missing paragraph page numbers to help determine where the problem is.

4. Can you translate only a few pages from the scan?

Yes, you can. Use desktop client page operations to split the required pages and submit the translation separately. Split files are smaller and easier to verify.

📝 Summary

If the translation of the scanned document is empty or missing paragraphs, it is mostly due to issues with text recognition, page quality, or output options. Checking in the order of "whether the original text is selected, whether recognition is complete, whether the page is clear, and bilingual cross-checking" usually helps quickly identify the cause. After translating important content, it is still necessary to verify paragraph by paragraph with the original text.

Open OmniDoc, upload the document, and start translating