OCR app for Mac: turn images and PDFs into searchable text

ImageXtract is a native OCR app for Mac that turns screenshots, scans, photos, and PDF pages into structured, searchable, and exportable content. Import a source, let local OCR identify the text and regions on the page, then review the result, search across your documents, and export it as Markdown, JSON, HTML, or Word.

Apple Silicon-optimized inference · Local OCR · Searchable projects · No cloud OCR service required

Get ImageXtract on the Mac App Store →

Stop retyping what is already in an image

Extract text from images. Bring in screenshots, scans, receipts, invoices, notes, diagrams, and other image files, then turn their visible content into selectable text.

OCR entire PDFs or only the pages you need. Choose individual pages or use a page range before processing, with progress and cancellation for larger documents.

Keep the result useful after extraction. ImageXtract preserves detected regions such as titles, headings, paragraphs, tables, lists, captions, and figures so the output is more than a plain block of OCR text.

ImageXtract runs as a focused Mac document workspace: your sources live in projects, completed pages remain available between launches, and extracted content can be searched or exported whenever you need it.

OCR software for Mac that keeps documents local

Text inside an image is easy to see and surprisingly difficult to reuse. A screenshot may contain an error message, a scan may hold a signed form, and a PDF may be packed with tables that you need in another document. Copying that information by hand is slow, and ordinary PDF viewers do not always make scanned pages searchable.

ImageXtract is built for that gap. It combines image-to-text OCR, PDF text extraction, document organization, and structured export in one native macOS app. The workflow stays close to the source: import a file, inspect the page with its detected regions, find the content you need, and choose the format that fits the next step.

Extract text from images and screenshots on Mac

ImageXtract accepts common image formats including JPEG, PNG, WebP, TIFF, BMP, GIF, AVIF, HEIC, and HEIF. HEIC and HEIF images are converted locally before OCR, so they can follow the same extraction workflow as the other supported formats.

You can choose files with the picker, drag them into a project, or import an image directly from the clipboard. If you are looking for how to copy text from a screenshot on Mac or convert an image to text on Mac, this is the core workflow: bring the pixels into ImageXtract, run OCR locally, and copy the recognized result instead of retyping it. A dedicated Quick Extract area is useful when you need a fast answer without filing the source into a long-term project. Clipboard images and one-off screenshots can be processed, reviewed, and exported without interrupting the rest of your workspace.

For recurring work, projects provide a clearer home for related material. Create a project for research notes, invoices, product documentation, receipts, customer records, or anything else that benefits from a searchable collection of extracted pages. ImageXtract keeps the original file and its processed page together in the project hierarchy.

Review OCR results with page layout and region labels

OCR is more useful when you can see how the extracted text relates to the original page. ImageXtract opens completed pages in a result workspace with the source image or rendered PDF page on one side and an Extracted content inspector alongside it.

Color-coded bounding boxes show where each detected region appears on the page. Hovering a region in the inspector highlights the matching box on the canvas, and selecting a search result can take you directly to the corresponding page and annotation. That makes it easier to verify a number, find the right paragraph, or check whether a table was interpreted as expected.

Detected content can include titles, headings, paragraphs, tables, captions, lists, and images. Table regions receive a scrollable HTML preview, while detected image or figure regions can be viewed at full resolution. A text region can be copied with one click; an image region can be copied as a crop, which is useful when the visual itself matters as much as the surrounding text.

This layout-aware approach is especially helpful for documents where position carries meaning: invoices, forms, reports, lecture notes, slide screenshots, and pages with a mixture of text and graphics.

OCR PDFs on your Mac, page by page

ImageXtract treats a PDF as a document made of pages rather than as one opaque upload. When you open a PDF, a local preparation view shows page thumbnails and lets you select all pages, clear the selection, or enter ranges such as 1-4, 7, 9-12. For anyone comparing a PDF OCR app for Mac, that page-level control is useful when a long document only needs partial processing.

Pages are rendered locally and added to the extraction queue incrementally. OCR can begin as soon as rendered pages are available, while the interface keeps per-page Queued, Extracting, Extracted, and Failed states visible. You can cancel preparation, avoid processing pages that are already ready, and export all completed pages as one consolidated document.

For a long report, this means you can process the chapters or appendices you actually need instead of waiting for an entire file to be handled. ImageXtract also keeps PDF search scoped to the selected document when you are reviewing a specific file.

Password-protected PDFs are not currently supported, but ordinary PDFs and their rendered pages stay inside the same local project workflow as imported images.

Search extracted text like a document library

Once OCR is complete, the extracted content becomes searchable instead of remaining locked inside a screenshot or scan. Search can cover the selected project or the selected PDF, with results that include highlighted snippets and document and page context.

ImageXtract supports everyday searches as well as more precise queries. You can search all words, use prefix matching, look for an exact phrase such as "annual revenue", combine alternatives with OR, and exclude terms with a leading minus sign. Region, document, and page filters narrow the results further—for example, label:table, document:invoice, or page:12.

Clicking a result opens the matching document and page, with the relevant detected region selected when available. That turns OCR into a practical way to retrieve information from a growing local archive, rather than a one-time conversion step.

Export OCR results to Markdown, JSON, HTML, or Word

Different tasks need different kinds of output, so ImageXtract offers several export formats for pages, complete documents, and projects.

• Markdown creates an ordered representation of extracted regions, with links to exported image crops where available.
• JSON preserves region labels, normalized coordinates, recognized text, and crop pointers for structured workflows.
• Spatial text produces monospaced text that approximates the original page layout.
• Spatial HTML keeps positioned content, semantic headings, lists, tables, and images in a browser-friendly document.
• Word documents combine extracted text, tables, multiple pages, and embedded image regions in a .docx file.

Exports can be written to a folder or copied to the clipboard. Text formats are placed on the clipboard as text, HTML also includes native HTML clipboard content, and Word output can be copied as a file. Project export creates one output document per completed source and adds safe numeric suffixes when a name already exists.

That makes ImageXtract useful both as an OCR reader and as a bridge into the rest of your workflow: move a table into a report, send structured regions to a script, paste spatial HTML into a compatible editor, or open a multi-page result in Word.

A private, local OCR workflow

ImageXtract is designed for documents that should remain on your Mac. Its OCR stack is more than a thin wrapper around a remote API: the app bundles a quantized Baidu Unlimited OCR model inside a highly optimized C inference engine with Metal kernels for Apple Silicon. The runtime is built to get the most out of Mac hardware while keeping recognition local, responsive, and available offline.

The model, tokenizer, C runtime, and Metal resources ship with the app inside a sandboxed local worker. Recognition and search operate on local files, so a cloud OCR account is not required for the extraction workflow. Imported originals, rendered PDF pages, OCR text, annotations, the full-text search index, and exports are stored locally.

The app does not need to send document images to a remote service for recognition, which makes it a natural fit for private notes, internal records, screenshots, and work performed offline. Its release configuration disables model and image downloads during OCR, giving you a predictable inference pipeline that does not change depending on network access.

Built for projects that keep growing

ImageXtract is not limited to a single image at a time. Projects can contain images and PDFs, with a persistent hierarchy that makes it easy to return to a source and its individual pages later. Document lists, PDF page lists, and search results load in pages, while thumbnails, images, and SwiftUI views are rendered lazily to keep the interface responsive.

Extraction state is persistent as well. Pages move through queued, processing, done, and failed states, and interrupted work can be recovered when the app launches again. Processing uses one local OCR worker at a time to keep resource use predictable, with controls to pause, resume, restart, or inspect the latest worker error.

Those details matter when a project contains more than a handful of scans. You can build a local archive of reference material without giving up the ability to search, filter, review, and export one page or one document at a time.

How to use ImageXtract

1. Create a project, or open Quick Extract for a one-off clipboard image or file.
2. Add an image or PDF with the file picker, drag and drop, or clipboard import.
3. For a PDF, select the pages or ranges you want to process.
4. Choose the project’s Faster or Quality OCR profile when needed.
5. Open a completed page to review its boxes, extracted regions, and plain text.
6. Search the project or PDF, then export the page, document, or project in the format you need.

The app keeps the original source alongside its managed page artifacts, so returning to an earlier result does not require repeating the import step. If the document changes, you can process the updated source as a new item and preserve both versions in the project.

Who is ImageXtract for?

ImageXtract is for Mac users who regularly work with information that arrives as pixels instead of text: researchers digitizing papers, students extracting notes from lecture slides, developers searching screenshots and error messages, operations teams processing invoices and forms, and anyone turning scans into editable documents.

It is also useful when privacy or offline access matters. A local OCR app can keep sensitive reference material inside the Mac while still providing the search and export features people expect from a document tool. The project structure helps when the workflow is recurring; Quick Extract keeps it lightweight when the task is temporary.

If you need a simple way to extract text from an image on Mac, OCR a PDF locally, search scanned documents, or preserve the layout of a page for export, ImageXtract brings those jobs into one focused workspace.

Why ImageXtract is different from a basic image-to-text converter

Many image-to-text tools stop after returning a text box. ImageXtract continues with the parts that make extracted content useful: persistent projects, page-aware PDF processing, visual region review, searchable OCR results, layout-preserving formats, and exportable image crops.

That combination is the point. You can check where a value came from, search for it again later, copy the exact text or image region, and choose an output format that keeps the information usable outside the app. The OCR result becomes a working document rather than a disposable conversion.