Bulk OCR: images and scanned PDFs to text
Give it a list of image or PDF links. It reads each page with Tesseract and returns one result row per page: the text, a confidence score (Tesseract's own estimate, not a measure of accuracy), and optionally the lines with their positions.
What it does
- A status on every page:
ok,no_text,unreadable(confidence under your floor),too_bigorfetch_failedwith the reason. One bad file does not fail the run, and good pages are always returned. - Images (png, jpg, tiff, bmp, webp, gif) and PDFs; PDF pages are rendered at the resolution you choose and read as images.
- English, German, French, Spanish, Italian, Portuguese and Dutch, alone or combined.
- A cap on pages per run (default 200), so a big list cannot surprise you. Nothing is kept after the run except your results.
- Output as plain text, lines with confidence and boxes, or paragraphs as markdown.
- Callable by AI assistants through Apify's MCP server.
Good to know
This is OCR, not vision: it reads printed text. Handwriting and low-resolution scans read poorly, and there is no table or layout analysis. Every PDF page is rendered and read as an image; an existing text layer is not used. Pages are read one at a time on CPU, so a long run takes a while; file and page counts per run are capped (maxPages). Up to 100 MB per file; localhost and private addresses are refused. Check important results.
Price
US$0.004 per page read (US$4 per 1,000), Apify platform usage included; unreadable, empty and failed pages are not charged. Set maxPages low for a first test.
A bug or a feature you need? Open an issue on the Store page.
More tools from Venture Ace