Repolode
Submit
google-ai-edge

google-ai-edge/mediapipe

Cross-platform framework and ready-made ML solutions for running vision, audio and text models on-device across mobile, web, desktop and IoT.

View on GitHubWebsite
Stars
37,361
Forks
6,194
Open issues
331
Last commit
8 hours ago

What is it?

MediaPipe is an on-device machine learning toolkit from Google AI Edge. It provides MediaPipe Tasks, cross-platform APIs and libraries for deploying pre-trained models for tasks such as vision, audio and text, along with MediaPipe Studio for testing and benchmarking solutions in the browser.

At a lower level, MediaPipe Framework lets developers build perception pipelines as graphs of calculators that process streams of packets. It targets C++, Android and iOS, and the project covers mobile, web, desktop, edge and IoT deployment. Input data such as images, video and text is processed locally on the device.

The project is a good fit for developers who want to add camera or media-based ML features without sending data to a server, or who need a customizable streaming pipeline for real-time media processing.

Key features

  • 01Pre-trained vision, audio and text solutions via MediaPipe Tasks.
  • 02Cross-platform APIs for Android, iOS, web, Python and C++.
  • 03Graph-based framework built from calculators and packets for streaming pipelines.
  • 04On-device inference, with input data kept local to the device.
  • 05MediaPipe Studio for testing and benchmarking solutions in the browser.

Use cases

  • Adding real-time face, hand or pose tracking to a mobile app.
  • Running on-device object detection or classification in a web page.
  • Building a custom video or audio processing pipeline for live streams.
  • Deploying ML features on edge or IoT devices where data must stay local.

Quick start

  1. 1

    Pick a solution guide, such as vision or text, from the developer docs

  2. 2

    Follow the setup guide for your platform, such as Python or Android

Things to know

  • MediaPipe Tasks send performance and usage metrics to Google, and app developers must obtain user consent where required.

Topics

#machine-learning#on-device-ai#computer-vision#mobile#deep-learning#pipeline-framework#video-processing#inference#android#audio-processing

FAQ

What is mediapipe?
Cross-platform framework and ready-made ML solutions for running vision, audio and text models on-device across mobile, web, desktop and IoT. MediaPipe is an on-device machine learning toolkit from Google AI Edge. It provides MediaPipe Tasks, cross-platform APIs and libraries for deploying pre-trained models for tasks such as vision, audio and text, along with MediaPipe Studio for testing and benchmarking solutions in the browser.
Is mediapipe free to use? What license does it have?
mediapipe is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is mediapipe still maintained?
Actively maintained. The last commit to mediapipe was on Oct 22, 2026.
What can I use mediapipe for?
Adding real-time face, hand or pose tracking to a mobile app. Running on-device object detection or classification in a web page. Building a custom video or audio processing pipeline for live streams. Deploying ML features on edge or IoT devices where data must stay local.
How do I install mediapipe?
Pick a solution guide, such as vision or text, from the developer docs Follow the setup guide for your platform, such as Python or Android
Where can I get mediapipe?
The source code and installation instructions are on GitHub: https://github.com/google-ai-edge/mediapipe

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects