AI-powered OCR API for instant text & data extraction

We provide a 14 days free trial so you can fully test any of our OCR models — no credit card required. After this, we offer different pricing tiers depending of the volume of pages processed and the features you might need. See the pricing page for more information.

Yes, it is absolutely possible to batch process and convert PDFs with Mindee. The best approach depends on how your "batch" of images or documents is organized.

Here is how you can handle batch processing based on Mindee's architecture:

1. Multiple separate image files (e.g., a folder of JPGs or PNGs)

By default, Mindee's standard API endpoint processes one file per HTTP request. To batch convert a large folder of separate documents, you simply handle the orchestration on your client side:

• Concurrent API Calls: You can write a script (using Python, Node.js, etc.) to loop through your files and send multiple API requests concurrently.
• Asynchronous Endpoints: For enterprise-scale volumes, Mindee provides an asynchronous API (/predict_async). Instead of holding the connection open, you push your batch of PDFs into a processing queue and then use webhooks (or polling) to retrieve the structured JSON data as each file finishes processing.

2. Multiple documents clustered in a single file

If your "batch" is actually a single large file containing multiple different documents (like a 50-page PDF of mixed mail, or a single photograph containing four different receipts laid out on a table), Mindee has native AI tools specifically built for this:

• Auto-Crop: If you upload a single document that contains several distinct items, Mindee's Auto-crop feature can automatically detect, isolate, and crop each item into clean, individual file ready for data extraction.
• Auto-Split: If you upload a large multi-page batch scan, the API's intelligent boundary detection detects where each individual document begins and ends, automatically slicing the massive file into discrete, logical records.
• Auto-Classify: Once the batch is separated, the routing engine acts as a digital architect to instantly categorize each document by type (e.g., separating invoices from contracts) and sends them to the correct extraction pipeline.

By combining your own asynchronous loops for individual files with Mindee's built-in Crop, Split, and Classify features for grouped files, you can build a highly efficient, automated batch-processing pipeline!

The supported document formats are:

• PDF (application/pdf) – While technically a document format rather than an image, it is natively supported for all extraction models.
• JPEG / JPG (image/jpeg)
• PNG (image/png) – Must be non-animated.
• WebP (image/webp)
• TIFF / TIF (image/tiff) – Both single-page and multi-page TIFFs are supported and processed similarly to PDFs.
• HEIC (image/heic) – Apple's High-Efficiency Image Container format.

‍

📌 Important technical limitations to keep in mind:

To ensure the OCR pipeline runs smoothly and successfully returns your JSON, make sure your files adhere to the following constraints:

• Maximum file size: 100 MB per file.
• Maximum page count: Up to 200 pages per document.
• File state: The files cannot be encrypted, and PDFs must not be password-protected.

Mindee supports every alphabets, every languages, every handwritten human-readable documents.

‍

Most top-tier APIs now support handwriting and 100+ languages. However, expect a 15-20% drop in confidence for cursive handwriting compared to printed text. For niche languages, verify if the OCR engine supports the specific character set (e.g., Cyrillic or Arabic).

Mindee could be the best fit for you if you need a reliable API, to extract line items variations with high-level accuracy.

Accuracy varies significantly based on the layout. While "Key-Value Pairs" (like Total Amount or Date) are easy, Line Items (Description, Quantity, Unit Price) are the hardest to parse because every vendor uses a different table style.

‍

Benchmark Tip: Don't trust the marketing "99% accuracy" claim. Test the same set of 50 "messy" invoices across vendors to see who misses line items or confuses the "Quantity" with the "Tax Rate."

Getting JSON is step one; getting valid JSON is step two. Most modern APIs, like Mindee allows you to define a data schema. To ensure your database doesn't crash:

• Use Pydantic (Python) or Zod (TypeScript) to validate the API output.
• If the extraction doesn't meet the schema (e.g., a missing mandatory invoice_id), flag it for human review.