SKL-666666/image-analysis-skill

README 摘要

🖼️ image-analysis — Structured Image Analysis for Text-Only LLMs Give text-only AI models (DeepSeek, GLM, etc.) the ability to "see" images : OCR text extraction + shape/color/control/icon recognition + table recognition + layout & hierarchy analysis, output as a structured Markdown report . In one sentence: it translates an image into a text report a model can read — so text-only models can understand images too. Great for: UI screenshots, web pages, flowcharts, architecture diagrams, tables & reports, form pages, chat transcripts, text documents . For photos/posters it provides color palette and layout (limited content understanding). ✨ Features Capability Description 🔤 Dual-engine OCR Windows built-in OCR + RapidOCR run in parallel, de-duplicated and merged for higher accuracy 🔷 Shape recognition Rectangles / circles / lines / buttons / badges, with coordinates (%), colors, and guessed functions 🧱 Layout & hierarchy Region grouping + parent–child containment tree (which element lives in which card) 📊 Table recognition Grid tables (with vertical lines) + borderless tables (columns inferred from text positions) 🎨 Color character map Rendered with text color codes so models c…

分类