How to use pre-built models

Overview

SmartDocs AI Studio offers a wide range of pre-built models based on the various business documentations and user case.

In this example, we will show how to use one of pre-built models called WaterBill to extract business model/information from a new water bill.

📘

Note:

For a full list of supported pre-built model, please refer to section 'USE MODEL' Pre-built Model for details.

Create a new project

In this case, we will create a new project with project type 'Water Bill' which will use a pre-built model.

Firstly, click on the '+ New Project' button under the 'PROJECTS' section to create a new project and select 'Water Bill' as the project type.

This project will be shown in the project list once created and you can enter the project by double-clicking the project name.

The image in below shows an overview of project created above and because the project is using a pre-built model, there are no 'LABEL' and 'TRAIN' options in this project as the model is already trained.

Use the model

It is straight forward to use the model, just upload a new water bill by clicking on the 'Upload Files' button under 'USE' section.

The SmartDocs AI Studio will initiate document prediction/data extraction automatically once the file is uploaded. A spinning Tools icon indicates that the prediction/data extraction is in progress.

Prediction/Data Extraction in progress

The spinning Tools icon will become a folder icon once the process completes and user can review the result by clicking the documentation.

Prediction/Data Extraction is completed

The image below shows the results of prediction/data extraction, the model correctly predict the following values according to the annotations from pre-built model: Property Address, Due Data, Issue Date, Total Amount while the value of Client Address is masked in purpose.

🚧

Note:

The WaterBill model is pre-built based on a heap of real water bills from various service providers in the Australian market. As such, this model may not perform well in case it is used to predict/extract information from water bills from other countries/markets.