Holistic landmarks detection task guide
Stay organized with collections
Save and categorize content based on your preferences.
The MediaPipe Holistic Landmarker task lets you combine components of the
pose, face, and
hand landmarkers to create a complete landmarker for
the human body. You can use this task to analyze full-body gestures, poses, and
actions. This task uses a machine learning (ML) model on a continuous stream of
images. The task outputs a total of 543 landmarks (33 pose landmarks, 468 face
landmarks, and 21 hand landmarks per hand) in real-time.
An upgraded version of this MediaPipe Solution is coming soon! The MediaPipe Legacy
Solution for this task is available on
GitHub.
[[["Easy to understand","easyToUnderstand","thumb-up"],["Solved my problem","solvedMyProblem","thumb-up"],["Other","otherUp","thumb-up"]],[["Missing the information I need","missingTheInformationINeed","thumb-down"],["Too complicated / too many steps","tooComplicatedTooManySteps","thumb-down"],["Out of date","outOfDate","thumb-down"],["Samples / code issue","samplesCodeIssue","thumb-down"],["Other","otherDown","thumb-down"]],["Last updated 2026-05-28 UTC."],[],[]]
