docTR-wrapper

Description

Wrapper for docTR end-to-end text detection and recognition.

Input

The wrapper takes a VideoDocument with SWT TimeFrame annotations. The app specifically uses the representative TimePoint annotations from SWT v4 TimeFrame annotations to extract specific frames for OCR

docTR Structured Output

From the docTR documentation

The docTR model returns a Document object

Here is the typical Document layout:

Document(
  (pages): [Page(
    dimensions=(340, 600)
    (blocks): [Block(
      (lines): [Line(
        (words): [
          Word(value='No.', confidence=0.91),
          Word(value='RECEIPT', confidence=0.99),
          Word(value='DATE', confidence=0.96),
        ]
      )]
      (artefacts): []
    )]
  )]
)

The docTR wrapper preserves this structured information in the output MMIF by creating lapps Paragraph Sentence and Token annotations corresponding to the Block, Line, and Word from the docTR output.

User instruction

General user instructions for CLAMS apps is available at CLAMS Apps documentation.

System requirements

• Requires mmif-python[cv] for the VideoDocument helper functions
• Requires GPU to run at a reasonable speed

Configurable runtime parameter

For the full list of parameters, please refer to the app metadata from CLAMS App Directory or metadata.py file in this repository.