MediaPipe Tasks Vision

Worker Lab

Camera Idle

Device Input Resolution Model Resize Width Face Mode Model Delegate Worker FPS 60 Inference Pace

Face Face landmarks, blendshapes, transform matrices Pose Body landmarks and world landmarks Hands Two-hand landmarks and handedness Segmentation Selfie mask overlay

Runtime Tasks API

Threading 4 Workers

Input Live camera

Video FPS 0 FPS

Mode Parallel

Experiment Result

No experiment run yet.

Camera stopped

Enable camera access to run MediaPipe Tasks Vision face, pose, hands, and segmentation in separate workers.