OCR or Optical Character Recognition has been around for a long time. You'll see OCR heavily used as a method to extract data contained within images or PDFs of questionnaires and/or surveys. This approach is often implemented to avoid manual data entry, saving on both time and labor. Although a proven technology, traditional OCR does have limitations. 

Amazon Textract on the other hand, is OCR on steroids. Built and provided by Amazon, Amazon Textract, under the hood, leverages machine learning to provide a level of service that surpasses many existing OCR solutions. One area where Amazon Textract excels is in scanning documents that contain tabulated information, and tables of figures, et cetera. Amazon Textract will not only recognize the characters within a table, but will maintain and format the table in the output. It can do so, because underneath, it's using machine learning technology. 

As already mentioned, Amazon Textract uses a machine learning model which has been trained on millions of documents, enabling it to detect and read special formatting features contained within any document. Complex layouts and formats used with many form-based documents can easily be detected, and therefore rendered back into the OCR-ed format. For example, financial, medical and/or tax-based forms tend to leverage many formats and structures to layout the questionnaire fields. Amazon Textract can intelligently detect these structures, extract and record them, and then encode them as key value pairs. Amazon Textract performs OCR on documents in either PNG, JPEG, or PDF formats. Documents can be submitted both synchronously or asynchronously, with APIs provided to support both methods. Uploaded documents into the Amazon Textract service can be deleted, and regardless, ownership of any uploaded document remains with the AWS account holder, and as such cannot be used anytime afterwards without their express consent. 

Amazon Textract has been seamlessly integrated into other AWS services, such as Amazon S3, AMS Lambda, AWS Batch and Amazon Elasticsearch Service. This allows you to quickly rollout solutions that encompass search and find features across your corpus of scanned documents. When working with Amazon Textract you can use the Amazon Textract console, or either the AWS CLI, or Java or Python SDKs. Over time, additional SDKs will be upgraded to provide Textract support. 

So, let's quickly summarize the key features that distinguish Amazon Textract as an OCR service. Machine Learning technology is used behind the scenes. Key-Value Pair and Table Extraction possible, with results encoded in JSON format. Bounding Box coordinates are provided, which can be referenced to determine exactly where the data was detected and extracted from. Confidence Thresholds are provided as a measure of confidence in terms of what was recognized and translated. If you're in need of an OCR solution that raises the bar and turns up performance, feature sets, and is both low-cost and pay-as-you-go, then Amazon Textract could be the right service.