perclos

A real time, webcam based, driver attention state detection/monitoring system in Python3 using OpenCV and Mediapipe

移动端实时疲劳驾驶检测

Real-time driver drowsiness and distraction detection system. EU regulation C(2023)4523 compliant gaze zone detection.

Local driver monitoring, deterministic fatigue alerts, and inspectable session history built with Python, FastAPI, OpenCV, and MediaPipe.

Edge Facial Behavior Analysis for Astronaut Multimodal Digital Twin — Winner of ASI Space Hackathon BEX2026 (Zero-Frame Retention, NASA NSBRI PERCLOS & Cardiovascular Twin)

Real-time driver fatigue detection system built with Python, OpenCV and MediaPipe. Detects drowsiness using EAR, MAR, PERCLOS, head pose estimation and a fatigue scoring system with Streamlit dashboard.

Real-time driver drowsiness & distraction detection on a plain laptop CPU - four chained OpenVINO models + a 30s PERCLOS scorer. 25-30 FPS, <8 MB models, fully offline. No GPU, no cloud.

Detector de somnolencia - Ultralytics RTDETR

Real-time driver drowsiness detection using PERCLOS and microsleep analysis over time, not per-frame classification. YOLOv8 and MediaPipe compared on the same decision layer.

Real-Time Driver Drowsiness Detection with CBAM-CNN

Offline in-browser drowsiness monitor — PERCLOS eye-closure detection, runs entirely on-device

Real-time AI-based driver drowsiness and behavior monitoring system using computer vision, sensor fusion, and multi-level alerts.

Reproducible evaluation of an eye-state + temporal-alerting drowsiness pipeline, with subject-disjoint testing, baselines, failure analysis and oracle decomposition.

Real-time driver drowsiness detection with per-driver calibration, a simulated driver for offline evaluation, a trained classifier benchmarked against the rule-based detector, and a human-in-the-loop safety supervisor.

A.W.A.K.E. 2.0 — Raspberry Pi drowsiness detection with eye tracking, pan/tilt servo control, and alarm system

Real-time driver drowsiness detection: CNN eye-state classifier + facial landmark geometry, fused over time with PERCLOS. 98.48% on unseen subjects.

Real-time driver drowsiness detection system based on OpenCV, MediaPipe Face Mesh, and Raspberry Pi 5

Real-time driver drowsiness detection system using MediaPipe, EAR, MAR, PERCLOS, head-pose estimation, and intelligent alerting.

Real-time facial analysis from a webcam: emotion, drowsiness, and a pain proxy. Re-measured on subject-disjoint splits after the leaked-split numbers were withdrawn, where EAR and PERCLOS beat the drowsiness CNN.

A modular Python engineering scaffold for 2-DOF PID gimbal face tracking and real-time facial state monitoring (e.g., eye-closure/blink detection). Features hardware mocking, headless testing, and Raspberry Pi readiness.