Text Detection and OCR with Microsoft Cognitive Services

2022-03-28 · Source: Adrian Rosebrock, Author at PyImageSearch · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Software Development & Engineering · Depth: Intermediate, long

Summary

This tutorial details how to implement text detection and Optical Character Recognition (OCR) using the Microsoft Cognitive Services (MCS) API, part of Microsoft Azure. It is the second in a three-part series on text detection and OCR, following Amazon Rekognition and preceding Google Cloud Vision API. The guide covers obtaining MCS API keys, configuring a Python development environment with OpenCV and Azure Computer Vision libraries, and structuring a project with a configuration file for API credentials. A Python script is provided to make calls to the MCS OCR API, process image data, and display OCR results, including bounding box annotations. The MCS API demonstrates robust performance, accurately OCR'ing text from various challenging images, such as warning signs, low-quality bus timetables, and pixelated text, even handling rotated text bounding boxes.

Key takeaway

For AI Engineers evaluating cloud OCR solutions, consider Microsoft Cognitive Services (MCS) OCR API, especially if your projects involve low-quality or challenging images. While its implementation might be slightly more complex than Amazon Rekognition, MCS demonstrates strong accuracy in diverse scenarios. If you are already within the Azure ecosystem, staying with MCS could streamline your workflow, despite the polling mechanism for results.

Key insights

Microsoft Cognitive Services OCR API offers robust text detection, even on challenging, low-quality images.

Principles

• Cloud OCR APIs simplify text extraction.
• API key management is crucial for access.

Method

Obtain MCS API keys, configure Python environment with OpenCV and Azure Computer Vision, create a config file for credentials, then use a Python script to send images to the MCS OCR API, poll for results, and annotate output.

In practice

• Use `pip install opencv-contrib-python` for OpenCV.
• Store API keys in a dedicated configuration file.
• Implement polling for asynchronous API results.

Topics

• Optical Character Recognition
• Microsoft Cognitive Services
• Azure Computer Vision API
• Text Detection
• Python Programming

Best for: Machine Learning Engineer, AI Engineer, Software Engineer

Related on AIssential

• Text Detection and OCR with Amazon Rekognition API — Amazon Rekognition API offers highly accurate cloud-based OCR, providing line or word-level text detection.
• Text Detection and OCR with Google Cloud Vision API — The Google Cloud Vision API offers high-accuracy OCR with minimal code, requiring a JSON key for authentication.
• GLM-OCR (9B) - Local OCR Test | OCR, Document Extraction, Table Recognition — GM OCR is a small, MIT-licensed, two-stage open-source model excelling in complex document and custom data extraction.
• Surya OCR 2: The Lightweight Open Source Model That Is Redefining Document Intelligence — A lightweight, open-source vision language model can achieve top-tier document intelligence performance.
• I Tested 5 OCR Models on 6 Real-World Datasets — Here’s Which One You Should Actually Use — No single OCR model is universally best; performance varies significantly by text type and use case.
• I Spent May Evaluating Different Engines for OCR — OCR is a routing problem; no single engine excels across all document types.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Adrian Rosebrock, Author at PyImageSearch.