Use caseArchive and full-text search
HowOCR an image book list into searchable text and export Markdown into your library
ValueFind any book by search later instead of flipping through images
Xiaohongshu image text OCR tool
Pull the value out of Xiaohongshu image cards, all at once
Image notes increasingly bake the steps, book lists, prices, and specs into image cards, leaving the body with a line or two. You can see the text but can't copy it. MediaClaw's built-in image text extraction (OCR) reads the cover and every image, writes the text back to the collected record, and carries it through copy and export.
OCR the cover and every image
An image note often has a dozen-plus images counting the cover. Image text extraction reads every one of them, not just a single image, pulling the steps, book lists, prices, and specs into copyable text.
Download free
Results write back to the record
A record that only had title, body, and engagement becomes complete raw data once image text is added. You read it directly, copy it in full, and carry the image copy through CSV/Markdown export.
Download free
Raw data lives in the extension, ready to use
This raw data is built inside the extension itself, collected, recognized, and exported in one place with no external tools. You can use it and export to CSV/Markdown right away. Collection and export are free, turn on OCR when you need it.
Download free
Why MediaClaw
Why not phone screenshots or online OCR
The three old paths each stall on something, batch, accuracy, or moving images by hand. The real problem is the same: image text is disconnected from the note data you're already collecting.
Batch, not one image at a time
Phone screenshot text grabs one image at a time and can't be merged with the note's title, body, and comments. Image text extraction reads the whole note and lands results in one record.
Paid model, more accurate
Free OCR is inconsistent; complex layouts, stylized fonts, and vertical text produce errors. One wrong character in a price, title, or spec means rework. Image text extraction uses a paid model that's steadier.
No moving images by hand
Online OCR sites make you save and upload each image yourself, back to the manual grind. Image text extraction reads the collected note directly and skips the shuffle.
What the extracted text is good for
Pulling image text out is only step one. It gets useful once it's in the collected record.
| Use case | How | Value |
|---|---|---|
| Archive and full-text search | OCR an image book list into searchable text and export Markdown into your library | Find any book by search later instead of flipping through images |
| Remix and breakdown | Pull a benchmark account's card-format steps and scripts into text | Analyze its structure and word choices line by line |
| Cover keyword research | Covers are a key place to seed keywords; OCR studies what viral covers wrote | Apply that cover keyword approach to your own topics |
| Complete both image and video | Image text via OCR, video voiceover via transcript | Covers the raw data for both image and video notes |
Use caseRemix and breakdown
HowPull a benchmark account's card-format steps and scripts into text
ValueAnalyze its structure and word choices line by line
Use caseCover keyword research
HowCovers are a key place to seed keywords; OCR studies what viral covers wrote
ValueApply that cover keyword approach to your own topics
Use caseComplete both image and video
HowImage text via OCR, video voiceover via transcript
ValueCovers the raw data for both image and video notes
FAQ
OCR the images. After collecting the note with MediaClaw, run image text extraction on it; the extension reads the cover and every image one by one and writes text back to the record for copy or export.
Yes. Image text extraction reads the cover plus every image in a note, not just one, and writes results back to the record as text that carries through CSV or Markdown export for full-copy and search.
Free online OCR is inconsistent; complex images, stylized fonts, and vertical text produce errors, one wrong character in a price, title, or spec means rework, and you still save and upload each image by hand. MediaClaw's image text extraction reads the collected note directly with a paid model for higher accuracy. It's the tradeoff between free-but-maybe-wrong and paid-but-more-accurate; pick the latter for sensitive content.
It writes back to the collected record inside the extension, alongside the note's title, body, and engagement. Run OCR on the record and the result lands right there, ready to copy or export to CSV/Markdown, with no external tools.
Pick an image note and run OCR once
Install MediaClaw for free, collect a few notes, run image text extraction on one image note, and see how accurately it reads the book list or price table before you batch-process your benchmark set.
Updated