vision-llm 开源项目推荐（142 个）

yigitkonur /api-llm-ocr

PDF to markdown using vision LLMs — tables, layouts, and structure preserved

OpenMOSS /MOSS-VL

An open-weight 11B model series for long-form and real-time video understanding

ahnafnafee /local-llm-pdf-ocr

Convert scanned PDFs and images into searchable documents with any OpenAI-compatible vision server. Run locally with a Web UI or CLI.

qiuyiwu1989-star /opendesign

通过 Playwright + Vision LLM 提取可复用 Web 设计令牌的开放标准。AI 就绪。

mazsola2k /ai-video-editor

AI Video Editor Pipeline with Vision LLM Models

ceodaniyal /free-llm-image-to-text

Free OCR powered by LLMs using OpenRouter — extract text from images with no API costs. Works with image URLs and Base64 inputs using free vision-capable models.

JamesDavid /FlipScan

Turn a phone flip-through video, photos, or a PDF of a book into a clean, proofread EPUB/PDF/Markdown — local vision-LLM digitization with a web GUI you can drive from your phone.

AIPythoner /pymidscene

PyMidscene - Midscene.js 的 Python SDK 实现 | AI 驱动的自然语言 UI 自动化，告别选择器，用中文描述即可操作。与官方缓存格式完全兼容。

ceodaniyal /local-llm-ocr-ollama

Free, offline OCR using local LLMs with Ollama. Convert images to text with vision-enabled models running entirely on your machine — no cloud, no API costs, full privacy.

nanofatdog /video-to-prompt

🎬 Extract AI prompts from video using Vision LLM (llama.cpp API) — Gradio WebUI + CLI

amitpatole /agent-vision

Eyes for AI coding agents 👁️ — render → perceive → report → fix → re-render. A machine-graded visual feedback loop (DOM/contrast/OCR-grounded + optional vision LLM) agents consume to self-correct before claiming done.

aidalinfo /extract-kit

Powerful PDF data extraction library powered by AI vision models. Transform PDFs into structured, validated data using TypeScript, Zod, and AI providers like Scaleway and Ollama.

vdamov /D2R-AI-Item-Tracker

AI-powered OCR for Diablo II: Resurrected - batch-extract item tooltips from screenshots using Vision LLMs (OpenAI, Groq, OpenRouter, LM Studio/Ollama). No Tesseract or EasyOCR needed.

farhan-beg /riddle-web

Tom Riddle's diary, in the browser. Write on screen, the diary drinks your ink and answers in handwritten script. BYOK, no backend secrets.

rowanlin-dev /video-notes-pipeline

🎬 视频 → 图文笔记流水线：B站 / 小红书视频 / 本地文件一键生成带截图、可点击时间戳的 Markdown + PDF 笔记。学科自适应模板 + 全覆盖抽帧 + OCR去重 + AI视觉打分精选 + 思维脑图，模型可切换（任何 OpenAI 兼容接口），可选入 ima 知识库。

mvmv1428 /deepcode-v4

Unlock Claude Code with DeepSeek V4. Get Anthropic's agent tools with 95% lower costs and local vision.

Rajskij /doc-dolphin

AI health companion: reads lab-report images with a vision LLM (Ollama) and streams the analysis live. React + Express + MongoDB.

gonzaloMorenoc /smartVisionQA

Proof-of-concept for automated visual testing using local vision LLMs via Ollama — no cloud, no API keys, fully on-premise.

barni007-pro /ollama_desktop_client

A feature-rich desktop GUI for Ollama with Vision, RAG, and JSON support.

YiannisStergiou /pdf-2-markdown

A powerful CLI tool that uses Vision LLMs to convert complex PDFs (math, tables, scanned docs) into clean, structured Markdown. Perfect for local RAG pipelines