Pull text out of any image — phone photos of receipts, screenshots of dense documents, scans of handwritten notes — and get back clean copyable text. ImageSuite runs Florence-2 (lucataco/florence-2-large on Replicate), Microsoft's vision-language model that handles printed text, handwriting, multiple languages, and busy backgrounds substantially better than classical OCR.
AI OCR is a Pro feature ($9/mo flat).
JPG, PNG, or WebP. Higher resolution = more accurate extraction, especially for small or stylized text.
Output is plain text — copy to clipboard or download as a .txt file. No images are returned; this tool extracts text, not annotated images.
Florence-2 (lucataco/florence-2-large on Replicate). It's a multi-modal vision model — handles printed text, screenshots, signs, receipts, and reasonable handwriting.
Yes — Florence-2 supports many languages out of the box, though English and major European languages are strongest. Right-to-left scripts (Arabic, Hebrew) may need manual re-ordering.
Convert the PDF to images first using PDF → Image, then run OCR on each page.