Creating a Classification Model with the Classifier Training Tool

Creating a classification model using the Classifier Training Tool shares similarities with web-based classification but requires a specific set of well-configured sample images. This article provides a step-by-step guide on how to create an effective classification model using the training tool.

Requirements for Training:

• Collect a minimum of 40 well-prepared training samples to achieve optimal results.
• Ensure your sample collection includes single-paged and multi-paged documents to enhance model accuracy.
• All training images must be either 100% PDFs or 100% TIFFs.
• For invoices and attachments, aim for around 200 preferred images, aligned with your incoming volume.
• For documents like tickets and statements, a suitable number of samples ranges from 45 to 50.
• Eliminate any blank or “garbage” pages from the samples to prevent classification issues.
• Gather 5-10 samples per vendor, with emphasis on common vendors, for each specific DocType you’re training.
• Ensure that all pages in the samples are properly oriented.

Creating Folders:

1. Unlike web-based classification, you don’t need batches. Create folders for each DocType you’re training (e.g., invoices, statements).
2. Name these folders EXACTLY as the corresponding DocTypes, respecting capitalization.
3. Create a separate folder for the class model itself.

Selecting the Parent Folder:

1. After setting up the folders, choose the folder where the class model will reside.
2. Select the parent folder that contains the DocType folders.

Building the Model:

1. The model will construct itself based on the provided samples.
2. The time required depends on your computer’s processing power.
3. For extensive sample sets, consider starting the process at the end of the day for optimal performance.

Uploading the Model:

1. Once the model is built, use the “Upload Model Tool” to transfer it to the customer’s system.
2. Enter the customer’s server IP and port details.
3. Click “Upload Model” to initiate the process.

Conclusion:

Creating a classification model through the Classifier Training Tool involves meticulous sample selection, folder organization, and model building. By following these steps, you can ensure a robust classification model that accurately categorizes various document types. This enhancement to your system contributes to improved efficiency and accurate data organization.

Was this article helpful?

0 out of 5 stars

5 Stars		0%
4 Stars		0%
3 Stars		0%
2 Stars		0%
1 Stars		0%

5 Submit

Please Share Your Feedback

How Can We Improve This Article?

Name

Email

Submit