How Unnish CCTV multi-face recognition marks attendance for multiple employees simultaneously from existing IP cameras — no dedicated hardware needed.
The End of Fingerprint Scanners in Bangladesh
For two decades, fingerprint scanners have been the default biometric solution. But anyone managing a factory floor or corporate office knows the truth: wet fingers fail, calloused hands fail, silicon replicas succeed. Queue times during shift changes stretch to 30+ minutes.
Unnish CCTV Multi-Face Recognition eliminates every one of these problems — and it's an industry first in Bangladesh.
How Unnish CCTV Multi-Face Recognition Works
Unlike basic facial detection (which simply locates a face), Unnish uses deep neural network-based facial recognition — identifying who the face belongs to with extremely high accuracy.
The Pipeline
- Detection: The camera feed identifies human faces in real-time using multi-scale convolutional neural networks
- Alignment: Faces are geometrically normalized — correcting for head tilt, angle, and distance
- Encoding: A 128-dimensional face embedding vector creates a unique mathematical fingerprint
- Matching: The embedding is compared against the employee database. High-confidence match triggers attendance logging
- Logging: Employee ID, timestamp, camera location, and confidence score written to the HRMS in real-time
The entire pipeline completes in under 200 milliseconds.
What Makes This Different From Other "Face Recognition"
Many HRMS vendors claim face recognition, but they mean a selfie-matching feature on a tablet. Unnish CCTV Multi-Face Detection is fundamentally different:
- Multiple employees detected simultaneously from a single camera frame — not one-by-one
- Works with your existing IP cameras (Hikvision, Dahua, etc.) — no dedicated hardware purchase
- Multi-angle detection in crowded environments like factory gates
- Employees simply walk past — no stopping, no touching, no queuing
Works Alongside Offline Mobile Punch
For employees who work remotely or in areas without cameras, Unnish offers Offline Mobile Punch:
- AI-verified selfie on the Android app
- GPS location tagged with every punch
- Works without internet — stores on-device, auto-syncs later
- Zero data loss even during extended outages
And Offline-First Terminals
Dedicated attendance terminals operate 100% independently during internet outages — essential for Bangladesh's infrastructure:
- Terminals store punches locally during load-shedding
- Automatic sync when connectivity resumes
- No duplicate or missed entries
- Real-time status monitoring from admin dashboard
Real Deployment Impact
Factory Gate Scenario: A garment factory in Gazipur deployed Unnish CCTV across 4 entry points. 2,500 workers clock in during shift changes — a process that took 45 minutes with fingerprint scanners now completes in under 5 minutes with zero queues.
Corporate Office: A financial services company in Gulshan uses Unnish for their 300-person office. Employees walk in and sit down — the CCTV system handles everything silently. No hardware at desks, no friction.
Integration with the Full HRMS
Face recognition data flows directly into the Unnish HRMS dashboard. Managers see real-time attendance, late arrivals, overtime, and analytics. Payroll is automatically calculated based on verified biometric data — no manual reconciliation needed.
The IoT Device Hub provides MQTT monitoring of all connected devices — RFID terminals, biometric readers, and CCTV bridges — with live heartbeat tracking and online/offline status.
Ready to implement this technology?
Consult with our engineering team to architect a custom solution.
Contact Engineering