Form Recognizer
-
Simplify Local file path of Training source for cross platform mobile use (Xamarin)
The API console struggles with the local file path definition and I don't always want to use embedded assets/resources in Xamarin projects. So is there a robust/consistent way to use local file path (say for Android/iOS) - and is the only option the local application data folder?
1 vote -
Train using URL sources other than Azure blobs??
This query may already be out there...apologies in advance...
Is there any roadmap/plan to add general URL sources for Training api other than azure blob containers?
(Have a separate question relating to local file paths)1 vote -
What's the best way to train a table without line? (See Pic)
I know Form Recognizer does not yet support labeling tables as tables. Suggestions please.
Please see pic1 for reference. Number of Line Items may vary at large.
And, What can we do if some text is not recognized by the OCR labeling tool see pic1 ?4 votes -
2 votes
-
Error Loading Application - An error occured while rendering the application. Please try again
after prediction in progress...
1 vote -
Method of confirming prediction as correct or incorrect
I uploaded 5 documents –– 2 batches of similar files. However, the model's confidence is very low. When I upload files though the model is working correctly. I'd love to be able to confirm the model if it were correct and automatically have it added to the training so that I don't have to manually tag each of the fields for that document.
3 votes -
The folder path for the container may not be working.
I could not get tagging to work with jpgs in a folder in the container and the same folder name in the project in the Docker image.
Is there a GitHub project for this?1 vote -
Old labeling tool vs the new one
The original forms labeling tool (VoTT) creates vott files and the new one (OCR Form Labeling Tool) uses fott files. I labeled and trained a bunch of forms using the original labeling tool. Is there a way for me to migrate those vott files to the fott file format?
It seems kind of strange you would switch labeling tools without providing any guidance on how to migrate their trained forms.
2 votes -
Label screen has spinning cursor, never completes
I pulled the latest Docker image on April 27th, on two separate computers and two separate subscriptions/tenants. I can create the connection and the project, but upon project save, it moves to the label screen and seems hung.
I have tried the hosted Docker image and the local one. Both act the same, as does my corporate and personal Azure accounts. Any idea as to what I am doing wrong?
1 vote -
Differentiate string entities
Hi, at the moment, labels can be string / date / number. Could be better to differentiate string types like person / address / ... This is possible in TextAnalyze service.
To improve our training, do Form Recognizer use these features to extract labels:
- NLP ? (words type adjective / verbs, ...)
- words before / after the label ?
- word case ? (1st capital letter / all upper-case / ...)
- text position in document ? (top right / center / ...)Thanks for info
1 vote -
Increase the total size of the training data set
The current total size of the training data set can be up to 500 pages. I'm planning to use this service for thousands of my customers that have multiple types of forms. Is the support of large training data sets something that is in your scope for the near future?
2 votes -
I can't make a project, but I don't know what it means.
I can't make a project, but I don't know what it means.
Both API keys and endpoints are correct.
The same work was done well last week.Let me know if anyone knows how to handle it.
1 voteIssue resolved. Please use Form Recognizer sample labeling tool try out site located here https://fott.azurewebsites.net/ or pull the latest release of the sample labeling tool container – https://docs.microsoft.com/en-us/azure/cognitive-services/form-recognizer/quickstarts/label-tool#set-up-the-sample-labeling-tool
-
Add splitbar to Form Labeler
If you use long tag names it exceeds the visible area. It would be nice to be able to expand the tag name area via a splitbar.
1 vote -
Use higher contrast highlighting in the Form Labeling tool
Using different shades of green for words that are selected vs just highlighted from OCR does not provide enough visual indication. For those of us that are color blind it's difficult to see the subtle difference. You might want to check out Microsoft's Accessibility Insights tool.
1 vote -
Add lasso selection functionality to Form Labeling tool
It would be really nice to have a "Lasso" selection functionality in the form labeling tool. I know I can hold the left mouse button down to highlight multiple words but there are some areas of forms (e.g. remarks) where we have several hundred words that have to be selected. They usually are in a rectangle shape so hence a lasso selection would work perfectly.
6 votes -
When is V2 Form Recognizer container going to be available?
So the the V2 Azure service is available now in preview but not the container. We want to be able to run on premise so this is quite important to us. Does anyone have a roadmap for this product? I haven't seen anything online indicating the status of the Form Recognizer product.
3 votes -
Return the document results in form order
It would be nice if the Get Custom Model and Get Analyze Form Result API's returned the fields/documentResults arrays in order in which the fields appear on the form (e.g. top/down, left/right). Currently Get Custom Model returns the fields in alphabetical order and Get Analyze Form Result returns them in a random order. My forms have a few hundred fields on them so it takes my consumers a while to figure out the form mappings.
1 vote -
In addition to bounding box, include area, height, and width with Analyze Layout result
The Analyze Layout API call currently provides the bounding box as a result for each line of text. While this information is useful, it always requires the consumer to perform additional processing to make it useful.
Consider directly providing the key calculations as part of the service result to reduce redundancy on the consuming side.
Provide the following values:
- Area of the bounding box
- Height of the bounding box
- Width of the bounding box
At a minimum, consider providing the height and the width as the area can be easily calculated from that.
The height of the bounding box can…
1 vote -
Add support for placing result layout analysis result directly into Azure Storage or Azure Queues
The current model for using the Analyze Layout service endpoint requires polling to retrieve the completed analysis result.
This model is inefficient and resource intensive on the application side. Consider a case when using Azure Functions: this would require multiple Function invocations to check if the analysis is completed.
This can be partially alleviated by incorporating a Shared Access Signature type functionality so that clients can directly query the Get Analyze Layout Result endpoint OR provide a parameter which allows the caller to specify an endpoint where the resultant JSON can be stored as a file.
For example, add two…
1 vote -
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
- Don't see your idea?