# google-ai-edge/mediapipe

Cross-platform, customizable ML solutions for live and streaming media.

Repository: https://github.com/google-ai-edge/mediapipe
Canonical: https://ross.abutalabs.com/products/mediapipe
Homepage: https://ai.google.dev/edge/mediapipe
Language: C++
License: Apache-2.0
License Family: permissive
Topics: mediapipe, c-plus-plus, computer-vision, deep-learning, android, video-processing, audio-processing, mobile-development, machine-learning, inference, graph-framework, graph-based, calculator, framework, pipeline-framework, stream-processing, perception
Last push: 2026-08-26T21:13:46+00:00

## Health v2 (maintenance only)
Score: 94/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 83, longevity 100
- inputs: {"age_days": 2638, "days_push": 7, "days_rel": 36, "gap_med": 50.5, "n_releases_24m": 11}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 36731, forks 6134 (observed 2026-08-28T04:12:01.241457+00:00)

## What it is
MediaPipe is Google's cross-platform framework for deploying on-device machine learning solutions for live and streaming media. It provides pre-trained models, cross-platform task APIs, and a graph-based pipeline framework for vision, text, and audio perception across mobile, web, desktop, and edge devices.

## Use cases
- detect faces and hand landmarks in a live camera feed
- run on-device image classification on Android and iOS
- add gesture recognition to a mobile app
- classify audio in real time on device
- build a streaming media ML pipeline in C++
- customize a pretrained model with my own data
- run object detection in a web app

## When to choose
- you need ready-to-run, cross-platform on-device ML for vision, text, or audio
- you want pre-trained models plus tooling to customize them
- you need low-latency inference on mobile, web, or edge hardware

## When to avoid
- you need server-side or cloud-scale training infrastructure
- you require custom model architectures outside the provided solutions
- you need a lightweight single-purpose library rather than a full framework

## Facets
- artifact type: framework
- maturity: stable
- function: machine-learning, computer-vision, image-processing, video-processing, audio-processing, llm-inference
- domain: machine-learning, computer-vision, mobile-development, cross-platform, artificial-intelligence
- platform: python, cpp, cross-platform, embedded
- tags: on-device-ml, pipeline-framework, pretrained-models, edge-ai, streaming-media, google, android, ios, web

## Member repositories
- google-ai-edge/mediapipe (main) score 94

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:12:01.241457+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T16:25:55.068572+00:00, confidence not recorded.
  - readme: https://github.com/google-ai-edge/mediapipe (fetched 2026-08-28T04:12:01.241457+00:00, sha 5a37c40d42f0)
  - homepage: https://ai.google.dev/edge/mediapipe (fetched 2026-08-29T07:47:22.983327+00:00, sha 1432ef7d17dc)
- Data as of 2026-08-30T08:39:29.467469+00:00.