YAP Pakistan
Production-grade AI-powered KYC solution that has successfully onboarded 100K+ users through automated identity verification, face liveness detection, and biometric face matching.
Healthcare & LogisticsGRAPH NEURAL NETlive2023
MyNatek
AI-based computer vision system that automatically monitors hospital cleaning activities in real time using object detection, human pose estimation, ST-GCN activity recognition, and multi-object tracking.
Key Metric
94%
Activity Recognition Accuracy
ST-GCN classification on held-out hospital test footage.
REF: ACTIVITY_RECOGNITION
MyNatek is an AI-powered hospital cleaning compliance platform that monitors hygiene activities in real time using existing CCTV infrastructure. The system detects cleaning staff, identifies their tools, tracks their movement patterns through hospital wards, and uses spatio-temporal graph convolutional networks (ST-GCN) to verify that required cleaning procedures are being executed correctly—without any manual observation or clipboards.
Hospital-acquired infections (HAIs) cause tens of thousands of preventable deaths each year. Manual supervision of cleaning staff is sporadic and subjective—supervisors can only be in one place at a time—leaving large windows where non-compliance goes undetected. Cleaning logs are paper-based and easily falsified, providing no real-time insight.
Pose & Tool Detection
PoseNet estimates 17-point body skeletons per frame; a parallel YOLO-based object detector identifies mop heads, spray bottles, and wipes. Multi-object tracking links identities across camera cuts.
ST-GCN Activity Classifier
Spatio-temporal graph convolutional networks model the joint skeleton as a graph and classify cleaning actions—mopping, wiping, spraying—over 3-second sliding windows, distinguishing genuine cleaning from false motion.
- 01.
Detects cleaning staff and their tools frame-by-frame using object detection and pose estimation on existing CCTV feeds.
- 02.
Classifies cleaning activities using ST-GCN, verifying whether prescribed protocols are being followed in each hospital zone.
- 03.
Generates live compliance dashboards and automated alerts for supervisors when required cleaning events are missed.
94%
Activity Recognition Accuracy
ST-GCN classification on held-out hospital test footage.
Real-Time
Compliance Monitoring
Live alerts to supervisors within 30 seconds of non-compliance.
Zero
Additional Hardware
Runs entirely on existing CCTV infrastructure.
