Processing Images with Mathpix Python SDK

This section explains how to send an image to the Mathpix API using mpxpy and retrieve structured OCR results.

Example: Processing an Image from URL

You can pass an image URL to the image_new() method to receive structured data output such as Mathpix Markdown (MMD) and individual OCR lines.

from mpxpy.mathpix_client import MathpixClient

client = MathpixClient(
    app_id="your-app-id",
    app_key="your-app-key"
)

# Process an image file
image = client.image_new(
    file_url="https://mathpix-ocr-examples.s3.amazonaws.com/cases_hw.jpg"
)

# Get the Mathpix Markdown (MMD) representation
mmd = image.mmd()
print(mmd)

# Get line-by-line OCR data
lines = image.lines_json()
print(lines)

The mmd() method returns a single string of LaTeX-style Mathpix Markdown.
The lines_json() method returns a list of objects, each representing a line of text or math, along with metadata such as coordinates, confidence scores, and handwriting detection.

Visual Example: From Image to Mathpix JSON

This is a real example that demonstrates how a handwritten equation is processed into structured output:

Example Output

[
  {
    "type": "math",
    "cnt": [[49, 332], [49, 0], [774, 0], [774, 332]],
    "included": true,
    "conversion_output": true,
    "is_printed": false,
    "is_handwritten": true,
    "id": "090b9d3a52e24421846eb864288ea20b",
    "text": "\\( f(x)=\\left\\{\\begin{array}{ll}x^{2} & \\text { if } x<0 \\\\ 2 x & \\text { if } x \\geq 0\\end{array}\\right. \\)",
    "confidence": 1,
    "confidence_rate": 1
  }
]

This data includes:

The type of line (math, text, etc.)
Bounding box (cnt)
Whether the line is handwritten or printed
The extracted LaTeX expression (text)
A confidence score from the OCR engine

You can also pass file_path for local images instead of file_url.
lines_json() is especially useful for building interfaces or performing advanced analysis of each recognized element.