Incorporate handling large blocks of text into existing OCR pipeline : Implementation Ticket

[bora-skylight]

Acceptance Criteria

• Find a way to determine if image has single line or multi-line text
• If multi-line text, run through paragraph sub-divider
• If single-line, skip paragraph sub-divider and run through regular model
• Write tests and integrate with the current pipeline: every image that goes through our pipeline, goes through the above