Field Models in Transkribus use AI to enhance layout recognition in historical documents. Unlike standard options, these trainable models can be customized to identify specific fields in your documents such as regions, marginalia, name fields etc.
How to Train a Field Model
Step 1: Preparing the Training Data in Transkribus Desk
Before you can train a Field Model, you need to prepare your training data. Import the historical documents you want to use for training into a specific collection in Transkribus Desk and then tag the structure of your documents as explained in this article.
Step 2: Training a Model in Transkribus
Step 2.1: Choosing Training Data
- Navigate to Transkribus models and click "Train New Model" and then select "Field Model".
- First, choose a collection where you will train your model.
- Then select the specific documents or pages within the collection you want to use as your training data.
Step 2.2: Tag Selection
Select the structural tags that you’ve prepared during the data preparation phase. These tags will be what the model learns to recognize.
Step 2.3: Validation Data
By default, Transkribus selects 10% of your dataset as validation data. We recommend sticking with this, but you also have the option to manually set specific pages as your validation data.
Step 2.4: Model Setup
- Model Name: Give your model a name.
- Description: Provide a brief description of what the model is for.
- Preview Thumbnail: Optionally, you can add an image that will serve as a preview thumbnail for your model.
- Advanced Options: You can set the number of training cycles. These cycles indicate how many times the model will go through the training data to learn and adjust. More cycles could result in a more accurate model but may also risk overfitting.
Step 2.5: Summary
Review all the settings and data you have inputted. Once everything looks good, proceed to start the training process.
How to Use the Trained Field Model
Step 1: Selecting Documents for Recognition
After your Field Model has been trained, go to your Transkribus Desk and select the documents or specific pages you want to recognize.
Step 2: Recognition Process
- Click on the "Recognise" button.
- Go to the top of the recognition section and choose "Fields."
- Under your private models, locate and select your newly trained Field Model.
And that’s it! Your Field Model will now be applied to recognize the layouts of the selected documents or pages.