EVI Analytics — video analytics module designed to detect, recognize, count, and track objects according to models.
It is recommended to update the system core before installing the video driver. |
Requirements for the Video analytics module
The video analytics system requires a video card. To operate in a virtual environment, the hypervisor must support hardware passthrough of PCI/PCI-E devices, including the video card. Compatibility with virtualization does not ensure stable operation of the video analytics module. |
A discrete graphics card is required for the module to function. The minimum configuration for operation is listed below:
| OS | GPU | SERIES | VRAM | DRIVER |
|---|
| Ubuntu 22.04 | NVIDIA | Quadro or RTX | 8 GB | 545 |
Example of installed core and driver versions
There may be problems installing the driver due to the Linux kernel. Examples of working options are shown below:
| OS | Kernel | Driver |
|---|
| Ubuntu 22.04.4 LTS | GNU/Linux 6.5.0-44-generic x86_64 | NVIDIA-SMI 545.29.06 |
| Ubuntu 22.04.4 LTS | GNU/Linux 6.5.0-45-generic x86_64 | NVIDIA-SMI 575.64.05 |
Manual installation of docker and drivers
It is recommended to use the latest driver available in the OS repository. |
To work with the video card, it is necessary to install the driver:
sudo apt install -yfqq nvidia-driver-575 |
The next step is to install docker for working with the analytics container and nvidia-docker2 for working with the graphics card:
sudo apt-get update
sudo apt install apt-transport-https ca-certificates curl software-properties-common |
curl https://get.docker.com | sh \
&& sudo systemctl --now enable docker
distribution=$(. /etc/os-release;echo $ID$VERSION_ID) \
&& curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add - \
&& curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list |
sudo apt-get update
sudo apt-get install docker-ce docker-compose nvidia-docker2 |
sudo systemctl restart docker |
Add a user to the docker group:
sudo usermod -aG docker $(whoami) |
A server reboot is required. |
It is necessary to check whether the driver is running in the system after installation:
An error when executing the nvidia-smi command indicates that the video driver failed to start and/or there is no discrete graphics card. Video analytics cannot be launched without a working NVIDIA driver. |
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 575.144.03 Driver Version: 575.144.03 CUDA Version: 12.4 |
|-----------------------------------------+------------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 NVIDIA GeForce RTX 3080 Off | 00000000:01:00.0 Off | N/A |
| 0% 41C P2 113W / 370W | 3902MiB / 10240MiB | 5% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| 0 N/A N/A 2144100 C tritonserver 1296MiB |
| 0 N/A N/A 2144155 C tritonserver 1296MiB |
| 0 N/A N/A 2144163 C tritonserver 1296MiB |
+-----------------------------------------------------------------------------------------+ |
Configuring database access for container connection
Before starting the configuration, it is necessary to stop the EVI services.
sudo systemctl stop evi-core.service evi-scud.service evi-live.service evi-archive.service evi-analyzer.service |
It is necessary to open access for connecting the analytics container to the database.
/etc/postgresql/17/main/pg_hba.conf |
Add the following line to the IPv4 local connections section and save the file.
Since containers get IP addresses from the 172.16.0.0/12 subnet by default, it is necessary to specify it.
# IPv4 local connections:
host all all 127.0.0.1/32 scram-sha-256
host all all 172.16.0.0/12 md5 |
Open the database configuration file. Remove the hash symbol (#) to activate the listen_addresses parameter and specify the server's IP address in it. Then save the changes to the file.
/etc/postgresql/17/main/postgresql.conf |
#------------------------------------------------------------------------------
# CONNECTIONS AND AUTHENTICATION
#------------------------------------------------------------------------------
# - Connection Settings -
#listen_addresses = 'localhost' # what IP address(es) to listen on; |
#------------------------------------------------------------------------------
# CONNECTIONS AND AUTHENTICATION
#------------------------------------------------------------------------------
# - Connection Settings -
listen_addresses = 'localhost, 192.168.1.10' # what IP address(es) to listen on; |
To apply the changes, restart the postgresql database service:
sudo systemctl restart postgresql |
sudo systemctl start evi-core.service evi-scud.service evi-live.service evi-archive.service evi-analyzer.service |
Loading the image and starting the evi-analytics container
Analytics images have a large file size, so they may take some time to load. |
After configuring access to the database, it is necessary to upload the evi-analytics_1.4.0.sh script to the analytics server:
wget --no-check-certificate https://archive.eltex-co.ru/evi-raw/evi-1.4.0/evi-analytics_1.4.0.sh |
bash evi-analytics_1.4.0.sh EFNRS_DB_HOST="IP ADDRESS OF YOUR DATABASE" |
bash evi-analytics_1.4.0.sh EFNRS_DB_HOST="192.168.1.10" |
Performance verification
docker ps --format "table {{.ID}}\t{{.Names}}\t{{.Status}}" |
CONTAINER ID NAMES STATUS
f01deee7d6a5 user-triton1-1 Up 23 hours
d70d9e7004ac user-triton3-1 Up 23 hours
c011105320e9 user-triton2-1 Up 23 hours
d17d6261648f user-evi-analytics-haproxy-1 Up 23 hours
fa2db6812f3b user-evi-analytics-1 Up 23 hours |