MediaPipe
On-device ML framework and ready-made tasks for vision, audio, and text
What it is
MediaPipe is Google's framework for building machine learning pipelines that run on phones, browsers, desktops, and edge devices. It ships prebuilt tasks with pretrained models for things like object detection, hand and pose tracking, face landmarks, audio and text classification, and on-device LLM inference, plus a lower-level graph framework of packets and calculators for custom pipelines. App developers on Android, iOS, web, and Python use it to add real-time perception features that process camera, audio, or text input locally.