🖨️ Self-hosted scanned document processor with AI-powered naming. Receives scans via FTP, SFTP, SCP, SMB or a watched folder, reads them with a local Ollama vision model, gives each file a dated, descriptive name, and delivers it to WebDAV (Nextcloud, ownCloud), an SMB share, or a local folder.
Open-source flipbook.page clone — every page is an AI-generated illustration, click anywhere to explore deeper. Next.js + FastAPI + Modal. BYO keys.
Self-hosted AI photo intelligence tool. Uses local vision models via Ollama to describe, tag, rename, and search your photos. No cloud needed.
AI video editing skill that watches your footage before cutting. Vision model analyzes every frame, scores usability, drops bad takes, reorders by narrative, then renders with ffmpeg. For Claude Code, Cola, OpenClaw. 会看画面的 AI 自动剪辑 Skill——先用视觉模型看懂素材再决定剪哪几秒。
AI-powered browser automation agent using a dual-LLM architecture. The orchestrator (qwen3-vl-32k) creates execution plans from screenshots, while the executor (llama3.1:8b) translates steps into browser actions using an accessibility tree for reliable element selection. Local, private, powered by Ollama.
evMLP: An Efficient Event-Driven MLP Architecture for Vision
This is a fork of SpaceInvaderOne's repo to fix some issues I had with the software until he pulls the changes or fixes them himself. It also allows for integration to my gallery app Eyeris. github.com/vonhex/eyeris
One key press, one action, one card. A lightweight Windows 11 tray assistant behind the Copilot key that looks at your screen and helps, with cloud or local (Ollama) models.
Windows 本地 AI 自动壁纸：三级筛选（本地预筛/免费模型/付费复核）+ 探索-个性化加权排序，带离线可复现评估（成对 0.672 / NDCG@5 0.783 vs 随机 0.525）。626 个测试。
Self-hosted TikTok capture that reads what is inside the post: photo carousels slide by slide with their audio, video with its ASR captions, interpreted by a vision model you host into typed repo, prompt and guide entries. Android share sheet with a durable offline outbox. Django + SQLite, one container.
基于视觉大模型的前端(React+Antd)全自动 CRUD 代码生成器 🚀 / AI-powered full-automatic CRUD code generator.
Codex plugin that lets Codex read images via a pure vision model (base64 -> vision API -> back to main model). 让 Codex 通过纯视觉模型查看图片的插件：图片转 base64 后调用视觉模型，识别结果返回给主模型。
微信聊天截图 → AI 生成 4 条候选回复 → 点一条自动填进输入框（只填不发）。Windows / macOS
A hands-on GenAI and Agentic AI engineering project work demonstrating LLM applications, tool calling, structured outputs, RAG, embeddings, local models, evaluation, benchmarking, and production-oriented workflows.
A .NET command-line tool that uses a local Ollama vision model to generate descriptions and suggested filenames for images in bulk.
Self-hosted image & video content moderation API powered by any vision LLM. Go stdlib only (FFmpeg for video), OpenAI-compatible, 15 categories, async jobs, webhooks, Prometheus.
Native Windows Computer Use and configurable vision tools for DeepSeek Harness.
AI 桌面直播弹幕姬 — 让 AI 一边看你的屏幕一边像直播观众一样发弹幕吐槽
Offline AI diagnostic assistant for marine/industrial equipment — Upload PDF manuals, ask questions, get engineering-grade answers with citations. Vision-powered diagram understanding. 100% local: Ollama LLM + ChromaDB + Streamlit.
PDF ingestion, local vision-model reading, and Salesforce storage. A local ETL pipeline 🌤️
To associate your repository with the vision-model topic, visit your repo's landing page and select "manage topics."