Skip to main content
Project Overview

What TDI Built for Ha-Meem

TDI modernized security infrastructure at Ha-Meem Group’s C-TPAT and Packing Area by layering AI intelligence onto the existing surveillance network. The system identifies authorized staff through facial recognition, flags unauthorized entrants in real time, and detects critical safety hazards like smoke and fire. It replaces paper registers and memory based checks with a centralized, compliance ready digital platform connected to local alarms and WhatsApp alerts for instant response.

Problem & Solution

Challenges & What TDI Built

01

Manual Access Checks

Security teams were relying on visual memory, photo displays, and paper registers.

02

High Daily Movement

Two sensitive zones saw heavy traffic that required faster verification and auditability.

03

AI Safety Layer

TDI added facial recognition, intrusion detection, hazard alerts, and digital reports.

The Challenges

Guards manually matched faces against photo displays and paper registers

160 to 170 people moved through two zones daily, overwhelming a 22 person security team

No uniform policy made visual department categorization unreliable

No automation existed for unauthorized access, tailgating, loitering, or hazard detection

No structured audit trail left the facility dependent on constant human vigilance

What TDI Built

AI facial recognition cross referenced against an authorized personnel database

Real time intrusion detection for unauthorized entry, tailgating, and loitering

Integrated smoke and fire hazard monitoring inside live camera feeds

Instant WhatsApp Business API alerts with snapshot, classification, and timestamp

Automated CSV and SQL audit reports for compliance and administrative review

Measurable Results

Business Impact

Seconds not hours

Security breaches are flagged in real time instead of being discovered retrospectively

15-20 FPS per stream

High throughput processing enables fluid tracking across high traffic corridors

Compliance ready

Tamper resistant digital audit trails are generated for buyer inspections and internal review

Scalable foundation

Architecture is prepared to expand across the full 1,300 camera network in later phases

Engineering

Core Technologies

Software
PythonFastAPIRTSP ProcessingCSV / SQL Reports
AI & Vision
Facial RecognitionIntrusion DetectionHazard DetectionIR Low Light Optimization
Hardware
NVIDIA JetsonDahua IP CamerasUniview IP CamerasNVR Systems
Network
High Speed LANRTSP Multi StreamEdge to Cloud Migration
Alerts
WhatsApp Business APILocal Alarm IntegrationSnapshot Alerts
Dashboard
Monitoring DashboardEvent LogsAudit Reports
Who Built This

Team Involvement

2
AI / CV Engineers
2
Backend Engineers
1
Frontend Developer
1
Business Manager
1
Product Manager
3
Data Annotators
1
Customer Success
1
Security Consultant
Project Timeline

Development Phases

Phase 01

Planning & Infrastructure

Finalized software architecture, AI pipeline, and team structure. Established remote IP access to the factory NVR after data collection hurdles, then locked the project scope to facial recognition after pivoting away from floor compliance.

Phase 02

Proof of Concept

Demonstrated a working PoC tracking selected individuals, which led the client to approve full scale personnel coverage for the protected zones.

Phase 03

Hardware Setup & Data Ingestion

Installed and calibrated dedicated cameras at the Packing Area entrance with NVR firmware updates. Captured the first facial dataset of 69 authorized personnel for real world model training.

Phase 04

Scaling & Cloud Deployment

Scaled the AI model with newly acquired datasets, migrated from local edge processing toward a cloud based architecture, completed A and B Unit coverage, and shifted the deliverable toward an interactive monitoring dashboard based on client feedback.

Currently Active
Platform Capabilities

Core Features

Automated Access Control

Replaces manual security checks with database linked facial recognition that identifies authorized and unauthorized personnel instantly.

Real Time Intrusion Detection

Continuously monitors restricted zones to flag unauthorized entrants, tailgaters, and loiterers as they happen.

Integrated Hazard Monitoring

Analyzes live video feeds for early smoke and fire signatures embedded directly within the surveillance pipeline.

Instant Response Linkage

Connects AI detections to local physical alarms and WhatsApp Business API alerts with snapshot and classification tags.

Automated Digital Audit Trails

Generates compliant CSV and SQL reports of entry and hazard events for administrative and buyer review.

What Is Next

Future Prospects

01

Backend expansion to manage the full 1,300 camera network with concurrent RTSP stream processing

02

Phased rollout replacing paper based and memory reliant security protocols across the facility

03

Advanced threat analytics including crowd density mapping, anomalous behavior detection, and perimeter breach alerts

04

Integration with HR and payroll systems for shift based access control tied to employee work schedules

05

Unified security operations dashboard connecting all zones, floors, and buildings into a single intelligent monitoring platform