You can now directly integrate your MS SQL databases
You can head over to the External Integrations Tab and configure your MS SQL database. Once this is done, the data extracted from the files that you upload to Nanonets will directly flow into your MS SQL database. This includes labels as well as line items.
You can now select the exact page in the pdf you want to process through your model settings section.
For eg: Let's say you've uploaded a 5 page pdf form and the data that you're looking to extract is only on page 2, you can specify this in your model settings. This should reduce your API usage as well an speed up the processing time
Searching for files on the annotate section of your model is now much easier. The search supports filters as well as full text search on the text in the document.
You can add filters on the labels as well as find files where the annotation does not exist.
You can now add your own post processing and validations to your model via python code.
You can head over to the Manage Labels section of your model and get this done. It is available as the "Python Script" option under post processing and validations
This is a significantly powerful feature that lets you modify or validate the model's outputs exactly as per your specification
You can now add multiple conditions to match the data derived from a document in a database.
You can now setup your custom approval flows on all your models
The Extract date from text post processing should now be significantly faster and save upto 1-2 seconds in API response time if set on your model
The Derive field from database post processing and the Match databases validation have both been updated to also support fuzzy matching. The fuzzy matching runs a LIKE query on your database
For eg: Let's say the data extracted from your document was Apple and the Vendor name on your invoice was Apple Inc, Nanonets would now be able to match these data points and associate them.
Now derive a new field from your database, based on a field extracted from your image.
Set up post-processing on any field on your document using the Operation 'Derive field from database'. Specify the table and columns to match from and choose which column to derive the field from.
Now capture tables of any type and only extract the cells you want. The tables can have dynamic structures and the model will learn the data you want to extract. It will learn to extract only the data you need and ignore all the others.
Multiple cells can be extracted from the same row or column leading to more flexibility creating the most powerful table extraction tool in the world.
It's super easy to annotate and train for the data you want to extract. Draw boxes around any text of interest even if it has no clearly defined boundaries and Nanonets will learn to extract that information
