Skip to main content

MediaPipe Installation

MediaPipe is an open-source, cross-platform machine learning framework developed by Google Research, focusing on real-time multimedia processing.

MediaPipe Features​

  • Cross-Platform Support

    Supports Android, iOS, Linux, macOS, Windows, and other platforms with a unified API, reducing the workload for platform adaptation.

  • Efficient Real-Time Processing

    Optimized for low latency, suitable for real-time applications, and utilizes hardware acceleration (such as GPU, DSP) to enhance performance.

  • Pre-built Solutions

    Offers various out-of-the-box machine learning models and processing pipelines, including face detection, hand tracking, pose estimation, 3D object detection, and more.

  • Modularity and Extensibility

    Users can customize processing pipelines by combining different calculators and data flow graphs.

    Supports integration with custom TensorFlow, TFLite models, or other ML models.

  • Lightweight

    Models are optimized for mobile devices, balancing accuracy and computational resource consumption.

Installing MediaPipe​

You can install MediaPipe using the pip command.

tip

Before installing MediaPipe, it's recommended to have Python 3 (preferably Python 3.9+) and OpenCV installed. For OpenCV installation, refer to OpenCV Installation.

Currently, MediaPipe only supports Python 3.9 to Python 3.12.

radxa@device$
pip3 install mediapipe

Verifying MediaPipe​

Open a terminal and enter the Python interactive mode by typing python3.

radxa@device$
python3

After entering the Python interactive mode, enter the following commands to verify that MediaPipe is installed correctly.

radxa@device$
import mediapipe
print(mediapipe.__version__)

If the MediaPipe version number is successfully displayed, the installation was successful.