Azure AI Form Recognizer – An Overview
Azure AI Form Recognizer helps Identify and extract text, key/value pairs and table data from documents, such as, Business Cards, receipts, etc. This automation also comprises a Training Model activity that allows users to create their own custom models by training with his/her inputs and whose model will be stored in a location which can be accessed anytime.
Prerequisites
Step 1: If You need an Azure subscription, create a free account before you begin. If you are a subscribed user, log into your account to commence the process.
Step 2: Create a Cognitive Services Resource using your subscribed azure account. Follow the below link to create the cognitive resource.
Create Cognitive Services Resource
Step 3: You will get an Endpoint and two account keys as displayed below:
Sample Endpoint – https://westus2.api.cognitive.microsoft.com/
Sample Account Key – “1b*******395******4ae*********f7”
Prerequisites for Activities
All the activities will only run inside the Azure AI Scope activity and it accepts the above created endpoint and account key as input parameters.
The Following activities will run with the Azure Prebuilt Models:
- Analyze Business Cards
- Analyze Layouts
- Analyze Receipts
Train Model activity allows users to create their own custom models by training with his/her inputs and whose model will be stored in a location which can be accessed anytime.
For training a model, you will need a set of training data in Azure Storage Blob as shown below:
Once it is created, get the Shared Access Signature (SAS) URI by right clicking on the container name. The SAS URI will be the input for training model. Below is a sample SAS URI.
FAQs
1. What types of documents does Form Recognizer work with?
Form Recognizer supports printed and handwritten forms, PDFs and Images.
Input requirements
Form Recognizer works on input documents that meet these requirements:
- Format must be JPG, PNG, PDF (text or scanned), or TIFF. Text-embedded PDFs are best because there's no possibility of error in character extraction and location
- File size must be less than 50 MB
- Image dimensions must be between 50 x 50 pixels and 10000 x 10000 pixels
- PDF dimensions must be at most 17 x 17 inches, corresponding to Legal or A3 paper sizes and smaller
- For PDF and TIFF, only the first 200 pages are processed (with a free tier subscription, only the first two pages are processed)
- The total size of the training data set must be 500 pages or less
- If your PDFs are password-locked, you must remove the lock before submitting them
- If scanned from paper documents, forms should be high-quality scans
- Text must use the Latin alphabet (English characters)
- For unsupervised learning (without labeled data), data must contain keys and values
- For unsupervised learning (without labeled data), keys must appear above or to the left of the values; they can't appear below or to the right.
Form Recognizer doesn't currently support these types of input data:
- Complex tables (nested tables, merged headers or cells, and so on)
- Checkboxes or radio buttons.
2. What Languages does Form Recognizer Support?
Form Recognizer is available in English, with additional language availability growing.
3. What is the difference between Custom and Pre-built features?
The custom option uses five samples to learn the structure of your forms and intelligently extract text and data tailored to your documents. The prebuilt option does not use samples—the model has already been trained and is ready to use off the shelf to extract key fields from specific document types. The two prebuilt APIs currently available are Prebuilt Business Card and Prebuilt Receipt.
4. What is the layout Feature of Form Recognizer?
Layout extracts text using high-definition optical character recognition (OCR) tailored for documents. It also extracts the structures of tables (row and column numbers).
For Advanced Users
To know more on Form Recognizer language support refer the following link:
https://docs.microsoft.com/en-us/azure/cognitive-services/form-recognizer/language-support