google-ai-edge/mediapipe
Google's toolkit for running face, hand, pose and object recognition AI directly on your phone or browser, no cloud needed
MediaPipe is Google's open-source toolkit for building on-device machine learning features like face, hand, and pose tracking that run on phones, web browsers, desktops, and edge devices. It ships pre-trained models plus tools to customize them with your own data and visualize results in the browser. Because processing happens on the device itself, input data like images or video never has to be sent to a server.
What it does
- MediaPipe Solutions bundles ready-to-use AI 'Tasks' (APIs) for vision, text, and audio jobs such as object detection, text classification, and audio classification, deployable across Android, iOS, web, and desktop.
- MediaPipe Model Maker lets developers retrain these models on their own data, and MediaPipe Studio lets them visualize and benchmark results directly in a browser.
- Underneath the pre-built solutions sits MediaPipe Framework, a lower-level C++ system built from 'Packets' (data units), 'Graphs' (pipeline structure), and 'Calculators' (processing nodes) for building custom real-time ML pipelines.
- Because input data such as images, video, or text is processed on-device rather than sent to Google servers, apps can use these APIs even for privacy-sensitive data, though usage metrics about API performance are still sent to Google.
- The project has been used in real products including Google Meet's background effects, AR art features, sign-language interfaces, and prosthesis control, and the repository has over 36,000 GitHub stars.
Why it matters
MediaPipe gives developers a free, production-tested way to add real-time camera and audio AI features to apps without building models from scratch or relying on server round-trips. Its on-device privacy design and cross-platform reach (mobile, web, desktop, edge) make it a practical foundation for teams shipping perception features quickly.
Terms in this repo
- MediaPipe Tasks · 미리 만들어진 비전/텍스트/오디오 AI 기능을 앱에 바로 연결할 수 있는 API 묶음
- MediaPipe Framework · Packets, Graphs, Calculators로 실시간 처리 파이프라인을 직접 만드는 저수준 C++ 시스템
- on-device processing · 이미지나 영상 같은 입력 데이터를 서버로 보내지 않고 기기 안에서 바로 처리하는 방식
- Model Maker · 자신의 데이터로 기존 모델을 다시 학습시켜 맞춤화하는 도구
- MediaPipe Studio · 브라우저에서 모델 결과를 시각화하고 성능을 비교 평가하는 도구
Repository description (English)
Cross-platform, customizable ML solutions for live and streaming media.
Open on GitHubTrending repos
- vorssaint/vorssaint-utilsOne free menu bar app replaces a dozen paid Mac utilities
- Alishahryar1/free-claude-codeA local proxy that lets coding AI agents run on 49 free or cheap model providers instead of one paid service
- freestylefly/awesome-gpt-image-2A library of 532 reverse-engineered prompts that turn GPT-Image2 into a predictable image-making tool
- block/buzzAn open-source workspace where humans and AI agents chat, code, and review in the same rooms
- NousResearch/hermes-agentNous Research's Hermes is an AI agent that gets smarter the more you use it
- virgiliojr94/book-to-skillA tool that turns technical book PDFs into on-demand reference skills for AI coding agents
- VoltAgent/awesome-agent-skillsA single hub collecting over 1000 'how-to' manuals that make AI coding assistants act like experts
- anthropics/claude-plugins-communityA shared shelf where anyone's Claude add-ons get listed for install