System Requirements

The System Requirements document outlines the technical specifications and prerequisites necessary for the successful deployment and operation of the KloudVision video analytics platform. This document serves as a comprehensive guide for system administrators, IT personnel, and stakeholders involved in setting up and maintaining the KloudVision environment.

KloudController works with most x86 hardware, giving users the flexibility to choose hardware that suits their needs. While the base KloudController doesn’t require high-end specifications, certain services (like video processing or advanced displays) may need additional hardware capabilities..

Vision Controller Specifications

The specification of the vision controller is given below.

General Compatibility

Feature Details
Supported Systems Most x86 systems with recent Intel or AMD processors (e.g., Intel Core i3 with min. 11th Gen or above, or equivalent)
Virtual Machines Can run on a VM if the CPU supports advanced features like “host CPU mode”
Base Requirements CPU: 4 cores (vCPUs), RAM: 16 GB, Storage: 64 GB

Service-Specific Recommendations

Depending on the services you plan to run, you may need additional hardware capabilities:

For small setups (up to 5 cameras):

Feature Details
Supported Systems Most x86 systems with recent Intel or AMD processors (e.g., Intel Core i3 with min. 11th Gen or above, or equivalent)
Virtual Machines Can run on a VM if the CPU supports advanced features like “host CPU mode”
Base Requirements CPU: 4 cores (vCPUs), RAM: 16 GB, Storage: 64 GB
Small Setup (≤5 Cameras) Intel NUC with i7 11th Gen, 16 GB dual-channel RAM, integrated Iris Xe GPU, 64 GB storage, 1 Gbps Ethernet

For larger setups (More than 5 cameras):

• For Intel GPU:
Component Specification
GPU Type Intel GPU
CPU Intel Xeon with ≥ 4 cores @ 3.0+ GHz per camera
RAM 1 GB per camera
Storage Optional 100GB, only if video alert recording is enabled else base 256GB
Network 10. Gbps Ethernet (recommended for setups with more than 10 cameras)

Note:

  • Actual core-count depends on the model deployed on the controller, e.g., Crowd-Analytics might need only 3 cores per camera for 1080p resolution.
  • 4K feed might demand a core-count of up to 5.
  • FRS and other complex models might need up to 6 cores per camera.
• For NVIDIA GPU:
Feature Details
Supported GPU NVIDIA RTX-series, A-series, L-series, V-series, H-series or T-series GPU
Supported CPU Intel or AMD CPU with at least 8 cores
RAM Requirement Typically ≥16 GB (scalable depending on number of cameras and workload)
Storage Requirement Optional 100GB, only if video alert recording is enabled else base 256GB
Network Details 10 Gbps Ethernet (recommended for setups with more than 10 cameras)

Example Specifications

Here are simplified examples based on different scenarios:

Advanced Setup (KloudVision for 10 Cameras)

• Use Case: Large-scale analytics.

• Hardware:

• For Intel GPU:
Feature Details
CPU Intel CPU with 40 virtual cores total (4 cores per camera)
RAM 16 GB total (1 GB per camera + 4 GB base + 2 GB rounding off)
Storage Optional 100GB, only if video alert recording is enabled else base 256GB
Network 10 Gbps Ethernet

Note:

  • Actual core-count depends on the model deployed on the controller, e.g., crowd-analytics might need only 3 cores per camera for 1080p resolution.
  • 4K feed might demand a core-count of up to 5.
  • FRS and other complex models might need up to 6 cores per camera.
• For NVIDIA GPU:
Feature Details
CPU Intel or AMD CPU with 32 virtual cores total (0.8 vCPU per camera)
GPU 1 x NVIDIA L4 GPU
RAM 64 GB total (1 GB per camera + 4 GB base + 2 GB rounding off)
Storage Optional 100GB, only if video alert recording is enabled else base 256GB
Network 10 Gbps Ethernet

Camera Specifications

Note: POE compatible preferred.

Hik-Vision:

Specifications:

  1. Min 1080p* 1920 Camera resolution i.e, 2MP
  2. Support to RTSP url

URL structure for Hik-Vision:

rtsp://username:password@<IP_ADDRESS_COMES_HERE>/cam/realmonitor?channel=1&subtype=0

For Hik Vision cameras, both RTSP and HTTP URL’s work fine. There is no need to use HTTP streams as RTSP works fine and is lag-free.

AXIS cameras:

http://username:password@<IP_ADDRESS_COMES_HERE>/mjpg/1/video.mjpg

For AXIS cameras, RTSP feed lags behind by almost 1.5 minutes. Use HTTP only.