Image Translation Table

The Problem

When engineers need to extract text from UI screenshots, flowcharts, menus, posters, or scanned documents and translate it into English, manual transcription can lead to inconsistent terminology, duplicated effort, and difficult-to-review batches. This skill turns text in images into a structured Chinese-English Excel table that is easier to review, reuse, and hand off.

How It Works

The skill identifies text in an image, groups it by region or position, and produces a bilingual comparison table. Key steps include:
- Load glossary: Prefer translation_glossary.xlsx when available, then use glossary terms and translation memory.
- Read and classify: Extract visible text and organize it by UI region or document section.
- Match terms: Reuse existing glossary entries and memory hits to reduce inconsistent translations.
- Translate and check: Translate each item, then run a consistency check and mark mismatches with orange highlighting.
- Export Excel: Generate single-image or batch workbooks with Python + openpyxl; batch mode gives each image its own worksheet and adds a 00_notes sheet.

Limits and Notes

It fits text-centric images such as UI screenshots, flowcharts, architecture diagrams, menus, posters, marketing materials, and scanned documents. Blurry images, stylized typography, dense layouts, and brand-specific wording may still need human review. If the glossary is missing, term matching is skipped and a new glossary may be created after translation; output files are saved in the current working directory.