Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
davidcastagnetoa avatar

Mediapipe Face Mesh

  • 5 installs
  • 1 repo stars
  • Updated March 10, 2026
  • davidcastagnetoa/skills

Extracts 468 real-time facial landmarks with MediaPipe Face Mesh to drive blink, head-turn and expression challenges in active liveness.

About

Generates 468 3D face landmarks in real time as the basis for active liveness challenges like blink and head-turn detection. Use it whenever a challenge needs precise facial movement or expression tracking.

  • Key landmark indices for eyes, nose and mouth with refine_landmarks iris points
  • Run on a dedicated thread to avoid blocking the main pipeline

Mediapipe Face Mesh by the numbers

  • 5 all-time installs (skills.sh)
  • Ranked #1,598 of 2,064 Data Science & ML skills by installs in the Skillselion catalog
  • Data as of Jul 29, 2026 (Skillselion catalog sync)
npx skills add https://github.com/davidcastagnetoa/skills --skill mediapipe_face_mesh

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

Listed on Skillselion
Installs5
repo stars1
Last updatedMarch 10, 2026
Repositorydavidcastagnetoa/skills

What it does

Extracts 468 real-time facial landmarks with MediaPipe Face Mesh to drive blink, head-turn and expression challenges in active liveness.

Files

SKILL.mdMarkdownGitHub ↗

mediapipe_face_mesh

MediaPipe Face Mesh genera 468 landmarks 3D en tiempo real. Es la base de todos los challenges activos: detección de parpadeo, giro de cabeza y expresiones faciales.

When to use

Usar en todos los challenges de liveness activo que requieren detectar movimiento o expresión facial específica.

Instructions

1. Instalar: pip install mediapipe. 2. Inicializar: mp_face_mesh = mp.solutions.face_mesh.FaceMesh(static_image_mode=False, max_num_faces=1, refine_landmarks=True, min_detection_confidence=0.5, min_tracking_confidence=0.5). 3. Procesar frame: results = mp_face_mesh.process(rgb_frame). 4. Extraer landmarks: landmarks = results.multi_face_landmarks[0].landmark → lista de 468 puntos (x, y, z normalizados). 5. Índices clave para cada tarea: ojos (33, 133, 159, 145, 158, 153 para ojo izquierdo), nariz (1, 4), boca (61, 291, 13, 14). 6. Proyectar coordenadas 3D a 2D con la dimensión del frame para cálculos de posición.

Notes

  • refine_landmarks=True añade landmarks de iris (precisión extra para blink detection).
  • Ejecutar en hilo dedicado para no bloquear el pipeline principal.
  • Documentación oficial: https://google.github.io/mediapipe/solutions/face_mesh

Related skills

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.