#99th of 9 in ML frameworks and libraries
mediapipe
Cross-platform on-device ML pipelines and ready-made vision, text and audio tasks
Live demo ↗
(opens in a new tab)Documentation ↗
(opens in a new tab)Website ↗
(opens in a new tab)Repository on GitHub ↗
(opens in a new tab)
- Stars
- 37.3k
- License
- Apache-2.0
- Last commit
- Oct 2026
- Last release
- Oct 2026
- Language
- C++
34,063 on 24 Mar 202637,273 on 13 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
MediaPipe is Google's C++ framework for building on-device ML pipelines from packets, graphs and calculators. It also ships MediaPipe Solutions: Tasks APIs with pre-trained models for vision, text and audio, Model Maker for customizing models with your data, and MediaPipe Studio for browser-based evaluation. Targets are Android, iOS, web, desktop, edge devices and IoT.
Who it is for: Mobile, web and edge developers adding on-device vision, audio or text ML
Strengths
- Tasks APIs cover vision, text and audio with pre-trained models, targeting Android, iOS, web and Python
- Input data is processed on device and not sent to Google servers
- Model Maker customizes models on your data; Studio benchmarks them in the browser
- Low-level Framework lets you build custom C++ graphs from calculators
Weaknesses
- Tasks APIs send performance and usage metrics to Google; app developers must obtain user consent
- Several legacy solutions are no longer supported; code is kept as-is
- README marks MediaPipe Solutions as a preview, early release
- Primary docs live on a separate Google site; the README is mostly links
What it needs
- GPU optional
- Docker
- Models: MediaPipe pre-trained task models, custom models via Model Maker
Also in ML frameworks and libraries
See all 9| Rank | Project | Score |
|---|---|---|
| 1 | transformersModel-definition library for text, vision, audio and multimodal models | 93 out of 100 |
| 2 | pytorchPython tensor library with GPU support and tape-based autograd for deep learning | 90 out of 100 |
| 3 | ultralyticsPython library and CLI for training and running YOLO vision models | 81 out of 100 |
| #4 | tensorflowTensor and neural network framework with Python and C++ APIs | 80 out of 100 |
| #5 | pytorch-image-modelsPyTorch library of image models, pretrained weights and reference training scripts | 75 out of 100 |