Form Recognizer
Attention!
We have moved our Customer Feedback & Ideas for Azure Cognitive Services portal to the Azure Feedback Forum.
-
Allow one-time use keys or shared access signature for direct submission
At current, there is no built-in mechanism to support one-time use access keys for clients to submit files directly from the client endpoint.
Form Recognizer should incorporate similar functionality to Azure Storage Shared Access Signature keys or provide a mechanism of issuing one time use keys to allow clients to directly submit files to the server to bypass additional superficial server-side processing.
At current, the model requires the usage of some additional processing layer to inject the secret APIM key. This is very much a waste of resources and application throughput as it is merely a mechanism of securing the…
2 votes -
Auto recognise wich custom model to use.
Hi
Woult it be posible to auto detect wich custom model to use? In a normal invoicing flow you would have one big pile of different invoices and send Them to be recognizer.
Then it would sort Them in 2 piles one wher a model was found and one without.
Do you understand what i Mean by that?11 votesWith the v2.1 preview announced last week, we have added support for “Model Compose” to support this functionality
https://docs.microsoft.com/en-us/azure/cognitive-services/form-recognizer/whats-new#august-2020 -
analyze results - missing labels in the result (lavel, value - pairs)
I have trained the model using labels.
In the result, I dont get something likelabel 1, value
label 2, valueIs there any way to get this label and value pairs?
1 vote -
Data Type for labels
Ability to specify a data type for a label, and have the recognizer look for that data type's formatting when extracting text for that field.
1 vote -
Recognizing form regions
At the moment we don't have features like recognizing checkboxes, radio buttons, signatures, etc.
Can form recognizer recognize a region of the form, so we can cut that part out and put for further processing?
For example, I would like to know where on the document are the radio button questions, so that I know in which area I need to do custom processing.
6 votesHi! For radio boxes specifically, we’ve added support for selection marks in our v2.1 preview. For region specification more broadly, this is only “started” status.
https://docs.microsoft.com/en-us/azure/cognitive-services/form-recognizer/whats-new#august-2020
-
Separate hand written text from printed text
Occasionally, OCR puts together printed and handwritten text as a single tag. I would like them to be separate.
For example:
Compay Handwritten company name
Printed text is usually the label while the handwritten text is what we are looking for.
6 votes -
Ignore dotted lines
When analyzing our user feedback form, I realized that the dotted line is occasionally recognized either as "........" or "**********".
Can we train it to ignore those?
Here is an example of an empty form: https://www.ssw.com.au/ssw/standards/forms/SSWEvaluationSurvey.pdf
1 vote -
Reinforcement learning
whenever something is wrong I want the model to learn
8 votes -
Predefined invoice Models
at the moment I have no working protoype
I see "training a model" in docu
But why should I train a model for invoices, If language and location is dedected, invoices follows specific rules- Tax, tax id and so one. My expectiation is: seems to be a German Invoice, take fitting model to extract data1 vote -
Test Form in Azure Portal
The Azure Portal should have some testing UI, with uploading a pdf and start the request.
2 votes -
Allow different formats to be trained in the same model
Allow different formats to be trained in the same model. Currently, for each format, a new model is created.
5 votes -
Analyzing forms with Fiddler
This may be obvious to many with more experience but it took me half a day to figure it out. I'm using Fiddler instead of cURL and when I was trying to call the analyze API with a file in the Request Body I was getting a 415 Unsupported Media Type. The API does not support multipart/form-data content type which is how Fiddler defaults binary files in the Request Body. The way to work around this is after selecting an "Upload file..." in the Request Body is to remove the multipart tags from the request body and change the content-type…
1 vote -
Model Id is not maintained.
Train the data using the sample label tool.
List all custom model using .Net SDK method (GetCustomModelsAsync())
Trained Model id doesn't available in the result3 votes -
2 votes
-
Structure of table data
First off, I love this tool, the key value pairs is very powerful and really solves a business need.
Having the ability to label the table elements would take this offering to the next level. If you were able to tag columns, eg. Column 1, column 2 and then the tool provides a row from the table as object...
{"fields":
…[{ "Address": {
...
},
"Name": {
...
},
"Size": {
...
}},
{ "Address": {
...
},
"Name": {
...
},
"Size": {
...
}},
{ "Address": {
...
},
"Name": {
...
},
"Size": {
...
}}]
}3 votes -
Release date for V2 including local container support?
For planning purposes we need to know when V2 is planned to be released. We also need to use the local container version so we need to know if the release of that will be different than V2 GA. Ideally we would like to test the local V2 container prior to GA.
5 votesYou can keep track of new releases and features on the What’s New page: https://docs.microsoft.com/en-us/azure/cognitive-services/form-recognizer/whats-new
-
Label tool and form processing
We did some testing with the examples. Particularly we want to start with the recognition of invoices so we need only the payment information, amount, … We started with the .NET SDK example and the custom form recognizer model. But we are receiving to much unnecessary information. Instead we are using the sample labeling tool which helps us to tag the necessary fields. Is there a way to upload the data of the training of the labeling tool in the same way as the customer form recognizer does?
1 vote -
Cognitive Service Container with V2.0 Support
The containers provided are with V1.0 API and all limitations ( no Layout API, 4Mb Dataset for training, ...).
When containers will be updated for V2.0 API ?Thx
8 votes -
Roadmap
Are you able to disclose what the roadmap is for the Form Recognizer product? I'm particularly interested in using the Custom Model and Labelling Tool.
6 votesForm Recognizer v2.0 (preview) is available and includes the ability to label forms and train a custom model. To get started follow the Train with labels quickstart – https://docs.microsoft.com/en-us/azure/cognitive-services/form-recognizer/quickstarts/label-tool
See what’s new in Form Recognizer v2.0 (preview) –
https://docs.microsoft.com/en-us/azure/cognitive-services/form-recognizer/whats-new -
Fix .Net SDK error for Analyze method
I'm getting the following error when using the analyze method of the SDK:
AnalyzeResult result = await _formRecognizerClient.AnalyzeWithCustomModelAsync(new Guid(modelId), stream, contentType: "application/pdf");
In case of HTML form data, the multipart request must contain a document with a media type of - 'application/pdf', 'image/jpeg' or 'image/png'.
Error code: UnsupportedMediaType
Status code: 415The data in the stream is correct, it contains a small pdf file.
More code leading up to the error:
var file = HttpContext.Current.Request.Files.Count > 0 ? HttpContext.Current.Request.Files[0] : null;
if (file != null && file.ContentLength > 0)
{
…using (var stream = file.InputStream){
AnalyzeResult result = await _formRecognizerClient.AnalyzeWithCustomModelAsync(new2 votes
- Don't see your idea?