Built by Maven

MediaPipeAnnotator

Mediapipe annotator integrates with mediapipe for processing of video to extract face landmark data.

The annotator can save an annotated .mp4 file and annotated frame image annotations, and also saves instantaneous blendshape score annotations, i.e. facial features that can be used to determine facial expression.

Prerequisites

The following must already be installed on the same machine that LaBB-CAT runs on:
  • Python 3
  • Support for Python virtual environments (venv)

Support for Python 'venv' can be installed on Ubuntu-like systems with the command:
apt install python3.10-venv

Installation

NB: During installation, the annotator needs to be able to connect to the internet so that required Python packages and model files can be installed.

Once installed, the annotator should function without an internet connection.