OCR Document Self-learning includes preset features that you can call directly or enable for scenarios such as intelligent pre-annotation.
FormUIE
FormUIE is an information extraction model that uses deep learning and task pre-training to enable pre-annotation for unstructured tables and forms of various formats on a single page. You can use the visual guide on the platform to configure fields for extraction without training a model.
If you use high-quality images and configure the fields correctly, the estimated pre-annotation accuracy is 50% to 80%.
Limitations: A single call is limited to 20 fields, and the total input cannot exceed 500 characters.
该文章对您有帮助吗？