paddleocr Robot

PaddleOCR

PaddleOCR Service is a synchronous HTTP API for extracting text and document structure from images. Each request uploads a single image, the server runs inference, and the result is returned in the same response — no polling required.

-

Input

Drop or click to upload image
Single image file, max 5 MB
JPEG
JPG
PNG
WEBP
Upload a clear image of text for best results. Supports printed text, handwritten notes, receipts, documents, screenshots, and more.

Recognition Result

Upload an image and click Recognize Text to extract text content

PaddleOCR Service

PaddleOCR Service is a synchronous HTTP API for extracting text and document structure from images. Each request uploads a single image, the server runs inference, and the result is returned in the same response — no polling required.

Quick Start

  1. Get an API key (Console → API Keys).
  2. Upload an image to a recognition endpoint, e.g. POST /v1/ocr with Authorization: Bearer YOUR_API_KEY.
  3. Read the JSON response — recognized text, polygons, scores, or document structure are returned in the result field.
  4. (Optional) Use X-Request-Id to correlate the call with the response header X-Request-Id if you contact support.

Endpoints

Method Path Description
POST /v1/ocr Full OCR pipeline (text detection + recognition) on the uploaded image.
POST /v1/text-detection Detect text regions in the uploaded image; returns polygon boxes and scores.
POST /v1/text-recognition Recognize text in a single, pre-cropped text-line image.
POST /v1/structure Parse the uploaded image as a structured document (layout, tables, formulas), with optional Markdown output.

See API Reference for full endpoint documentation.

Parameters Overview

POST /v1/ocr

Parameter Required Default Description
image Yes — Image file to process (supported image type or PDF).

POST /v1/text-detection

Parameter Required Default Description
image Yes — Image file to process.

POST /v1/text-recognition

Parameter Required Default Description
image Yes — Pre-cropped single-line text image.

POST /v1/structure

Parameter Required Default Description
image Yes — Image file to process.
return_markdown No false When true, the response includes a markdown field with a Markdown rendering of the document.

Languages

The recognition model (PP-OCRv6_medium) is a single unified multilingual model whose dictionary natively covers 50 languages — including Japanese, Simplified and Traditional Chinese, English, and 46 Latin-script languages. No language selection parameter is required.

Error Responses

Errors return a JSON body and a non-2xx HTTP status code.

json Copy 
{
  "detail": "File too large (24117248 > 20971520 bytes)"
}
Status When it occurs
401 Unauthorized Missing Authorization header, or the provided API key is invalid.
413 Payload Too Large Uploaded file exceeds the maximum image size.
415 Unsupported Media Type Upload Content-Type is not an image type or application/pdf.
422 Unprocessable Entity Upload is empty, or the file cannot be decoded as an image.
500 Internal Server Error Inference failed or produced no result.
504 Gateway Timeout Inference exceeded the maximum allowed processing time.

Full error format and details are documented in API Reference.

Limits

Constraint Value
Maximum upload size 20 MiB
Maximum inference time per request 120 seconds
Accepted upload Content-Type image/*, application/pdf
Color handling Images are processed as RGB; alpha channels are flattened

Unlock the most affordable AI hosting

Run models at scale with our fully managed GPU infrastructure, delivering enterprise-grade uptime at the industry's best rates.

Contact Sales