An Azure service that turns documents into usable data. Previously known as Azure Form Recognizer.
Hi, here are some considerations:
- As an experiment, try to train the custom model without labels to see if you get better results
- Ensure your training data set follows the input requirements (pay attention to the additional tips as well)
- Improve your model accuracy by labeling additional forms and retraining to create a new model (consider adding the prediction file to the training set, label it, and train again)
- Try using the new Form Recognizer Studio (public preview) to see if you get better results
--- *Kindly Accept Answer if the information helps. Thanks.*