Calculator
AI video analytics server & capacity calculator
Set your camera count and modules. The servers, GPU, storage and network you need update instantly.
Number of cameras
cameras
10
50
100
200
300
500
Deployment
Hardware
Modules
Intrusion detection
Always on
On of 50 cameras
Face recognition
~1.5 GB VRAM
On of 50 cameras
Weapon detection
~1.2 GB VRAM
On of 50 cameras
Leak detection
~1.1 GB VRAM
On of 50 cameras
Smoke and fire
~1.1 GB VRAM
On of 50 cameras
Servers required
—
GPU servers
—
cameras max (capacity)
⚠ Load >85% — recommend +1 server
Infrastructure load
—
GPU / chip—
VRAM—
Cameras per server—
Power draw—
Infrastructure—
—
RAM (estimate)
—
storage / 30 days
—
RTSP network
—
active modules
Active modules
How the sizing works
What the numbers are based on
The same GPU servers run every module on the camera streams you already have. These are planning estimates; an engineer checks them against your camera list before anything is ordered.
~25
cameras per RTX 4090 server
The baseline for intrusion detection. Each extra module on a camera adds GPU memory (about 1.1–1.5 GB), so fewer cameras fit per server.
~2 Mbit/s
network per RTSP stream
By default, video is analysed on your own server and stays on site. Storage covers 30 days of event clips, not continuous recording.
1–3 days
from server to live detection
Point Vengeea at the RTSP address of each camera. No camera is replaced and nothing is rewired.
Real frame · east fence · intrusion
Real detection · gas station camera · gun identified
Real detection · factory camera · smoke identified