davidcastagnetoa/skills

ear_blink_detection

Detectar parpadeo natural midiendo Eye Aspect Ratio (EAR) frame a frame con landmarks faciales

First seen Mar 2, 2026

Installation

$ npx skills add davidcastagnetoa/skills --skill ear_blink_detection

Similar popular skills

Related neighbors and high-traction skills in the same topics — useful to compare before installing.

Also in this package

Other skills from davidcastagnetoa/skills · top by installs.

npx skills add davidcastagnetoa/skills

Browse all from davidcastagnetoa/skills

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Not declared
Cursor Not declared
Codex Not declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Not declared

Repository health

Stars 1
Default branch main
Open issues 0
Status Active

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 1,298 B
  • docs SUMMARY.md 121 B

History

  1. First seen on skills.sh
  2. First recorded snapshot · 12 installs

SKILL.md

earblinkdetection

El Eye Aspect Ratio (EAR) mide la apertura del ojo usando 6 landmarks faciales. Un parpadeo natural produce una caída rápida y recuperación del EAR. Las fotos no parpadean.

When to use

Usar como challenge principal de liveness activo: solicitar al usuario que parpadee 2 veces en 5 segundos.

Instructions

  1. Extraer landmarks del ojo usando MediaPipe Face Mesh.
  2. Para el ojo izquierdo usar landmarks: [33, 160, 158, 133, 153, 144].
  3. Calcular EAR: EAR = (||p2-p6|| + ||p3-p5||) / (2 * ||p1-p4||).
  4. Umbral de parpadeo: EAR_THRESHOLD = 0.25 (ojo cerrado si EAR < threshold).
  5. Detectar secuencia: EAR normal → cae por debajo del umbral → recupera → cuenta como 1 parpadeo.
  6. Requerir mínimo 2 parpadeos en ventana de 5 segundos.
  7. Validar velocidad del parpadeo: un parpadeo natural dura 150-400ms; más rápido o más lento es sospechoso.

Notes

  • Fórmula original: Soukupová & Čech (2016) "Real-Time Eye Blink Detection using Facial Landmarks".
  • EAR_THRESHOLD puede necesitar calibración por etnia/edad; considerar threshold adaptativo.