Usage of neural networks🔗

Part of the Eocortex video analysis modules uses neural networks.

Neural networks are presented as an additional installation package.

Starting from version 3.6 of Eocortex, two different packages can be used:

The Eocortex Neural Networks Standard package
The Eocortex Neural Networks Special package

Note

Eocortex Neural Networks Standard and Eocortex Neural Networks Special packages support simultaneous use of multiple video cards for analyzing video streams. However, correct work of the analytics is possible only when using video cards that are identical to each other by their technical properties.

Comparison of neural networks packages

Capabilities	Eocortex Neural Networks Standard	Eocortex Neural Networks Special
The first appearance of the neural network package in Eocortex	2.5	3.6
Operating systems	Windows 10 Windows Server 2016 Ubuntu 20.04, 22.04 Debian 10, 11 CentOS 7	Windows 10 Windows Server 2016 Ubuntu 20.04, 22.04 Debian 10, 11 CentOS 7
Silent installation on Windows systems	Yes	No
The ability of some modules to work without a discrete graphics card	Yes	No
Minimum requirements when using a discrete graphics card	A processor that supports AVX instructions; An NVIDIA video card (GPU) with the computation capacity index of at least 6.5 and with at least 4 Gb of memory; the parameters and performance of the video card must be similar or better than those of NVIDIA GTX 1650 Super model; Version of the video card driver at least 460; Swap file at least half of the total RAM size. If the package will be installed on a virtual machine, it may additionally be required to: Enable support for AVX instructions in the guest machine settings; Use GRID drivers for GPU virtualization.	A processor that supports AVX instructions; An NVIDIA graphics card (GPU) with an index of compute capability of at least 7.0 or higher and video memory capacity of at least 6 GB; herewith, other characteristics and performance of the graphics card must be no lower than that of NVIDIA RTX 3050 or NVIDIA RTX 2060 model. Version of the video card driver at least 460; Swap file at least half of the total RAM size. If the package will be installed on a virtual machine, it may additionally be required to: Enable support for AVX instructions in the guest machine settings; Use GRID drivers for GPU virtualization.
Advantages	X	Frames are processed faster
Advantages	X	The load on the graphics card has been significantly decreased, which allows processing many more cameras simultaneously
Consumption of resources of the central processing unit (CPU) and random access memory (RAM)	Does not depend on the used neural network package
	Depends on the number of cameras being processed
	Depends on the video analysis modules used
Consumption of resources of the graphics card (GPU)	Depends on the video analysis modules used
Technologies supported by CUDA	FP32	FP32, FP16
When the central processing unit (CPU) load is high	Video analysis frequency decreases	Analysis modules stop working
Video analytics
Counting People in Queue	Yes	Yes
Emergency Vehicle Detection	Yes	Yes
Face Mask Detection	Yes	Yes
Face Recognition (Complete)	Yes	Yes
Face Recognition (Light)	Yes	Yes
Fire and Smoke Detection	Yes	Yes
Fall Detection	Yes	Yes
Object Classification and Counting	Yes	Yes
Search for Objects	Yes	Yes
Tracking	Yes	Yes
Uniform Detection	Yes	Yes
Unique Visitor Counting	Yes	Yes