Case Study: Streamlining Manufacturing Security with Nvidia Tesla V100 GPUs

Overview:

A manufacturing plant, equipped with 100 cameras, sought to fortify its security measures and operational efficiency through a comprehensive Video Management System (VMS). The goal was to implement sophisticated video analytics ensuring compliance, safety, and incident response across the facility.

Challenges:

  • Managing feeds from 100 cameras simultaneously for various security and compliance purposes.
  • Ensuring adherence to safety protocols, fire detection, intrusion monitoring, and operational area management.
  • Efficiently allocating GPU resources to handle diverse video analytics requirements.
  • Cost-effective hardware selection for seamless and efficient operations.

Solution:

  • The implementation involved leveraging Nvidia Tesla V100 GPUs known for their computational prowess and adeptness in handling complex AI-driven tasks.
  • Use case distribution across cameras was meticulously mapped to optimize GPU utilization efficiently.
  • SOP compliance, fire detection, intrusion monitoring, and area management were allocated to Tesla V100 GPUs for optimal performance.
  • A singular setup of Tesla V100 GPUs was chosen to streamline management, reduce complexity, and meet the computational demands efficiently.

Use Case Distribution:

  • SOP Compliance: 80 cameras
  • Fire & Loss of Primary Containment Detection: 100 cameras
  • Vehicle Movement and Manpower Detection in No Entry Zone: 80 cameras
  • Pathway Obstruction Detection: 30 cameras
  • Monitoring Number of People in the Operational Area: 4 cameras in Dahej + 6 cameras in Uran
  • Vehicle Speed Monitoring and NPR: 15 cameras
  • Flare Monitoring: Only on Thermal Cameras (Edge Analytics)

Pricing:

  • Tesla V100 GPU: $8,000 per GPU (9 GPUs required) - Total cost for GPUs: $72,000
  • System Requirements (estimated at $10,000 per setup) - Total cost for system setups: $90,000

Implementation:

  • Tesla V100 GPUs efficiently managed multiple use cases assigned to different camera feeds using their 64 cores and high computational capacity.
  • The allocation of use cases based on load (light, medium, heavy) optimized GPU resource utilization, ensuring seamless real-time monitoring and compliance enforcement.
  • The VMS, powered by Tesla V100 GPUs, swiftly responded to security threats, compliance violations, and operational inefficiencies, ensuring accurate detection and rapid incident resolution.

Results:

  • The deployment of Tesla V100 GPUs streamlined video analytics across 100 cameras, ensuring real-time monitoring and compliance enforcement.
  • The system's high accuracy and reliability in identifying security breaches and compliance violations significantly bolstered the facility's security posture.
  • The consolidation of use cases onto Tesla V100 GPUs resulted in cost-effectiveness, while maintaining exceptional performance standards.

Conclusion:

The strategic implementation of Nvidia Tesla V100 GPUs in the VMS highlighted their pivotal role in fortifying security and compliance in a manufacturing setting. By concentrating use cases onto a singular high-performance GPU type, the facility achieved operational efficiency, cost-effectiveness, and enhanced security measures.