Mediapipe to Dlib Face Landmarks Conversion

This tool converts Mediapipe's 478 face landmarks into Dlib's 68 face landmarks.

I created this tool because many projects still rely on Dlib's 68 landmarks format, while others use Mediapipe's face detector. To enable code reuse between both sides, this tool provides a practical solution.

Works that Use This Conversion

• flame-head-tracker: https://github.com/PeizhiYan/flame-head-tracker

Install

pip install .

Usage

from mediapipe2dlib.mp2dlib import convert_landmarks_mediapipe_to_dlib

# Convert Mediapipe dense landmarks to Dlib sparse landmarks
lmks_mp2dlib = convert_landmarks_mediapipe_to_dlib(lmks_mp)

Example