Image to Text OCR
Pricing
Pay per event
Image to Text OCR
Extract text from public image and PDF URLs with page-level OCR results.
Pricing
Pay per event
Rating
0.0
(0)
Developer
Sam Towers
Maintained by CommunityActor stats
0
Bookmarked
2
Total users
1
Monthly active users
a day ago
Last modified
Categories
Share
This Actor accepts public image or PDF URLs and writes one normalized dataset item per successfully processed page.
Input is JSON with sources (or urls) as a string or array, plus optional maxFileSizeMB, maxPdfPages, dpi, ocrMode (fast or accurate), quality (draft, balanced, or high), skipEmpty, minConfidence, and includeLines. Downloads are streamed with bounded timeouts and file sizes. PDF text layers are preferred; scanned pages are rendered and OCRed.
Each output contains sourceUrl, pageNumber, text, confidence, ocrMode, sourceSha256, and lines (each line has text, confidence, and [left, top, right, bottom] bounding coordinates). Duplicate source-page outputs are suppressed within a run. Successful pages are charged with page-processed, or accurate-page-processed when accurate OCR is selected. Failed downloads and unreadable pages are skipped without a page charge.
The image/PDF and OCR libraries are declared in requirements.txt; the Docker image installs the Tesseract runtime.
