Form Understanding in Noisy Scanned Documents

Functioning as derived from the FUNSD project for form understanding in noisy scanned documents. It was uploaded by user sijpapi to Hugging Face in September 2021. The specific row count, column structure, and file formats are not provided in the input.

Use Cases

• Train a model for semantic entity labeling on form documents, such as identifying 'questions' and 'answers'.
• Develop a layout analysis tool to detect and classify text regions like 'headers' or 'key-value pairs' in scanned forms.
• Build a relation extraction system to link related entities, such as connecting a 'question' field to its corresponding 'answer'.

Strengths

• Based on the established FUNSD benchmark for form understanding research.
• Dataset was uploaded to a major platform (Hugging Face) in September 2021.

Limitations

• The input provides no information on dataset size, number of documents, or specific annotations available.
• The description is a single URL with no details on data structure, features, or sample data.
• The dataset's recency is unknown beyond the 2021 upload date; underlying data may be older.

Provenance

Source

FUNSD (Form Understanding in Noisy Scanned Documents) project.

Freshness

Last updated on the platform was 2021-09-01.