Skip to workspace
PixelOkay

Document Intelligence Studio

Private document analysis workspace
New project
Tools
▤

Understand every page

Import a document to view its regions, text, tables, and structured extraction.

Ready to import
100%

Analyze documents locally, with reviewable results

PixelOkay Document Intelligence Studio helps inspect scanned images, digital PDF pages, text layout and ruled tables. Review visual regions, correct extracted content, compare pages, and export structured data. It is designed for local exploration rather than guaranteed legal, financial or high-stakes extraction accuracy.

Document layout inspection

Detect approximate headings, text blocks and visual regions. Where a PDF has selectable text, use its embedded text positions to identify paragraphs.

OCR and table review

Optionally download browser OCR tools and a language pack to recognize image text. Table grids are heuristic and require manual checking.

Compare and export

Compare page text and visible differences. Export JSON with normalized coordinates, CSV tables, page text and an editable local project.

EXPORT

Download results

Structured extraction excludes embedded images by default. Save a project if you need to reopen and edit regions.
DATA RECOVERY

Browser recovery copies

Browser backups are temporary, size-limited, and may be erased by browser storage cleanup. Download a project for durable storage. Damaged entries can still be downloaded before deletion.

Using Document Intelligence Studio

Import: Add PNG, JPEG, WebP or PDF. PDFs require an on-demand download of PDF.js. Scanned PDFs are rasterized; selectable text PDFs expose text-layer positions.

Detect: Detect layout suggests regions based on text or image contrast. Ruled tables may be detected from image lines; this is not AI-level semantic table understanding.

OCR: OCR downloads the Tesseract.js runtime and English model from third-party CDNs. OCR can be slow, and results may be wrong. Reviewer edits and filled tables are preserved during re-analysis. Cancel stops the ongoing session.

Edit: Draw on the page to add a region, or select a region to edit its type, bounds and text. Region coordinates are fractions of the page.

Compare: Select two imported pages to compare text and pixel appearance. Differences may reflect rescaling, noise or OCR errors.

Privacy: Page files are processed in your browser. External libraries and language files are fetched; user documents are not submitted to an OCR API. Browser backup is best effort, not permanent storage. Download an editable project to keep a reliable copy.

Using Document Intelligence Studio

Import supported image or PDF pages, inspect extracted text and review detected regions. Correct layouts or table cells and mark reviewed content before rerunning supported detection/OCR. Export the required text/data/image artifacts and keep editable project JSON.

Layout/table detection is heuristic and OCR needs human review; it does not guarantee complete document understanding. PDF imports download PDF.js 6.4.299, and optional OCR downloads Tesseract.js 7.0.0 plus worker/core/language assets. Document processing is browser-side, while dependency requests contact external providers. File/page/pixel/project budgets apply.

Workflow and output choices for Document Intelligence Studio

Inspect supported image or PDF pages, review extracted text and detected regions, correct table cells and mark reviewed content before exporting data or project files.

PDF imports request PDF.js 6.4.299 when that support is used. Optional OCR requests Tesseract.js 7.0.0 and worker, core and language assets. Processing runs in the browser, while those dependency downloads contact external providers.

Example workflow

  1. Open a supported image or PDF and select a page to inspect.
  2. Review text and suggested regions; correct a table or mark content as reviewed before rerunning supported detection or OCR.
  3. Export the needed text, table or image artifact and save project JSON to continue the review.

Authored example: Use a fictional one-page invoice with selectable text and one small table. Correct a suggested cell and mark it reviewed; this exercise does not claim OCR or table extraction accuracy.

Limits and handling: Region and table detection are heuristic, and OCR requires human review. The tool does not guarantee document understanding. File, page, pixel and project budgets apply; PDF/OCR dependency requests are external even though page processing is browser-side.

Document Intelligence Studio questions

Does the tool automatically understand or verify an entire document?

No. It helps inspect text and regions, but detection is heuristic and extracted content needs human review.

Does PDF or OCR processing make any network requests?

The page processing is browser-side, but PDF.js and optional Tesseract.js worker, core and language files are requested from external providers.

Related tools