Screen OCR Text Recognition

Problem

Screen text is often visible but not editable: it appears in screenshots, remote desktop windows, shared apps, or rendered document pages. ocr-screen targets this workflow by converting visible screen text into copyable plain text, especially for mixed Chinese and English content, table fragments, and batch capture tasks.

Core behavior

• Chinese and English recognition: supports Chinese and English text, which is useful for bilingual UI, annotations, field labels, and body copy.
• Table recognition: recognizes tabular structures in captured screens, helping keep table-like content easier to organize.
• Batch processing: handles multiple screen regions or items in sequence instead of one small crop at a time.
• Export-friendly output: returns recognized text as copyable plain text, with optional export to Word / Excel for downstream document or spreadsheet work.

Boundaries

• Accuracy depends on screen clarity, font size, contrast, and the integrity of table lines.
• It is oriented toward visible screen text; if the source file is natively editable, using the original text will usually be more reliable.
• Dependencies must be installed on first use, and the runtime environment can affect stable operation.