Spinovate - Watchdog for bad desk habits

About the Project

Spinovate is an innovative AI-powered posture monitoring system designed to combat the growing epidemic of poor desk posture. Built for the NCIT hackathon by Team Tech Tacticos, this system uses computer vision and machine learning to provide real-time feedback on multiple health parameters. The application analyzes head pitch angles, user distance from screen, and estimates critical medical angles including thoracic kyphosis, lumbar lordosis, T1 slope, and cervical spine anteversion - measurements typically calculated manually by orthopedists. With a Next.js frontend and FastAPI backend, Spinovate delivers physician-verified posture assessments alongside drowsiness detection, yawn monitoring, and blink tracking to ensure comprehensive desk health monitoring.

Key Features

• •Real-time posture monitoring with webcam using MediaPipe Face Mesh
• •Medical-grade spinal angle estimation (6 angles including thoracic kyphosis, lumbar lordosis, T1 slope)
• •Multi-parameter health tracking: head pitch, distance from screen, brightness monitoring
• •Drowsiness detection using Eye Aspect Ratio (EAR) with configurable thresholds
• •Yawn detection using Mouth Aspect Ratio (MAR) with frontal/non-frontal head position adjustment
• •Blink detection and tracking for eye strain prevention
• •Comprehensive alert system with severity levels (low, medium, high) and audio notifications
• •Session monitoring with detailed analytics and reporting
• •Real-time visual feedback with on-screen status table and pitch visualization
• •Physician-verified posture detection logic ensuring medical accuracy
• •Smart alert cooldown system to prevent notification spam

Challenges & Solutions

• ⚡Achieving accurate 3D pose estimation from 2D webcam feeds across varying lighting conditions
• ⚡Calibrating medical angle calculations to match orthopedist-level accuracy
• ⚡Balancing alert frequency to be actionable without causing alert fatigue
• ⚡Optimizing MediaPipe face mesh processing for real-time performance on standard hardware
• ⚡Implementing robust yaw angle detection for differentiating frontal vs. non-frontal head positions
• ⚡Designing an effective multi-threshold alert system for diverse posture issues
• ⚡Synchronizing frontend Next.js dashboard with FastAPI backend video stream processing

Outcomes & Impact

• ✓Successfully implemented physician-verified medical angle estimations comparable to manual orthopedic assessments
• ✓Achieved real-time processing with concurrent video streaming and metric calculation
• ✓Built comprehensive monitoring system tracking 6+ health parameters simultaneously
• ✓Implemented smart alert system with cooldown timers and stacked event tracking
• ✓Created production-ready FastAPI backend with professional service separation and error handling
• ✓Developed modern Next.js dashboard with real-time metric visualization and audio alerts
• ✓Won recognition at NCIT hackathon for Team Tech Tacticos