Event Security Monitoring
Event security monitoring solution with multiple SmartFace nodes each with 4 or more cameras and one central watchlist management system.
Main features
- High quality real-time face detection and person matching
- Instant notification of presence of unwanted persons
- Scalability - from small events to large events using the same solution
- Important Event Notifications
- Searchable History
Description
Event security monitoring solution powered by the SmartFace Platform allows to cover your events such as festivals, shows, gatherings in smart cities, shopping centers, public transportation, or any other public areas - ensuring that using real-time video analysis all the persons' faces are detected and matched against your watchlists. Allowing you to know the identity and location of people present through the whole event.
This includes the fast detection and notification of presence of people who’s access to the event has been restricted.
Additional Features
Complete Feature Set for Any Facial Recognition Use Case
We have designed the SmartFace Platform to be extremely fast while providing reliable facial recognition capabilities in real time.
Our proprietary facial recognition works precisely even on low-quality images or images made in subpar lighting conditions. SmartFace also accurately recognizes faces in different positions or partially obscured by hair, hands, or objects. SmartFace works well also in degraded lighting conditions or recognizes people wearing sunglasses and hats. It can subsequently track the movement of the subject across monitored areas.
Edge-to-Cloud Architecture
With our cascaded architecture, video streams are pre-processed on Edge cameras at each camera’s location, resulting in a dramatic reduction of bandwidth requirements and savings on server resources required at the central site. Such central site can be located within the premises, or due to minimal Edge stream network requirements it can be deployed in a cloud.
SmartFace can be easily scaled and utilize the resources of a server to its maximum.
Watchlists and Notifications
Our easy-to-use watchlist management helps prevent security incidents by swiftly blocklisting unwanted persons and sending alerts about their presence. Instant notifications for a detected and matched person or body minimize the time to action.
Monitoring a number of screens is exhausting, inefficient, and time-consuming. Our SmartFace notification mechanism allows the precise configuration of events and situations that will trigger a notification or an alert that are worth an operator’s attention. Alerts can be sent to other systems like smartphone applications or emails. These notifications can be fully integrated into 3rd party applications via excellent API coverage.
SmartFace can be used out of the box or can be integrated into your existing architecture and systems using an API. It comes with detailed product technical documentation and a well-documented API to ease integration.
Multiple Watchlists
Unlimited number of watchlists supports your custom security strategies and helps to monitor different groups of individuals according to their characteristics. You can add multiple images under one enrolled individual to learn different appearances of the individual over time. Ensuring exceptional accuracy even with a face mask on or when an individual’s face attributes gradually change over time.
Component Diagram
Please see below the components and their connection.
Hardware requirements
The Event Security Monitoring consists of 2 different parts. They differ in function and in hardware requirements:
- Central Node - one node to oversee general configuration and Watchlists with enrolled biometric profiles
- Local nodes - nodes for each location processing local cameras and presenting biometric notifications
Central Node
The Central Node acts as a Leader, is used to maintan the Watchlists and Watchlist members. Replicates the watchlists onto the Local Nodes.
Multi-site Watchlists Synchronization is based on Leader - Follower architecture that provides centralized watchlist management and increased resilience of SmartFace instances. For more information about Multi Site Deployment read a deployment guide.
In this setup, you can have one Leader SmartFace instance and multiple “Follower” SmartFace instances. Changes in Watchlists and Watchlist members are automatically synchronized from Leader to all Followers. This setup is especially useful when the SmartFace instances are geographically separated and are required to work independently even in case of connection errors.
Device Name | Description | Amount |
---|---|---|
NUC | A central SmartFace server to act as the source of truth and a central data hub Minimum requirements: Intel Core i7-1360P Raptor Lake 5 GHz, Intel Iris Xe Graphic, 8 GB Ram | 1x |
Switch | A central Switch to act as a unifying network architecture | 1x |
The Central Node, can be also running on a cloud. In that case, you can use the Amazon EC2 c7g.2xlarge instance.
Local Node
Processing of the Edge cameras happens on the Local Nodes. The Local Nodes act as Followers, replicate the watchlists from the Central Node.
Device Name | Description | Amount |
---|---|---|
Hanwha PND-A6081RV | Smart Camera supporting our SmartFace Embedded Stream Processor plugin, making it an Edge Stream camera Manufacturer’s link: https://hanwhavisionamerica.com/product/pnd-a6081rv/ | 4x |
POE Switch | Power Over Ethernet (POE) enabled 1GBit Switch with suggested 8 ports, while at least 4 ports have POE. | 1x |
NUC | A local SmartFace server to act as a local processing and security dashboard unit Minimum requirements: Intel Core i7-1360P Raptor Lake 5 GHz, Intel Iris Xe Graphics, 8 GB Ram | 1x |
Installation
Once you have the required hardware ready, you can start with installation. We recommend to mark each device, so it is clear what name does each device have and to which node it belongs.
Network Installation
Use the Component Diagram as a guide for how to plug different components into your network. The Edge cameras needs to be plugged into POE (Power over Ethernet) ports on your POE switches.
IP Address Configuration
After the devices are physically connected into the network we can set up the IP addresses for each device. Below is a table of IP addresses used for this package to make it run out of the box. On each device set the manual IP address as below and use the subnet mask: 255.255.255.0
and the default gateway: 192.168.17.1
.
NUC Nodes | IP Address | Description |
---|---|---|
Central (Leader SmartFace) | 192.168.17.10 | Central SmartFace Server |
#1 Local (Follower SmartFace) | 192.168.17.20 | Local SmartFace Server for the 1st Node |
#2 Local (Follower SmartFace) | 192.168.17.30 | Local SmartFace Server for the 2nd Node |
#X Local (Follower SmartFace) | 192.168.17.X0 | Local Cameras used within the Xth location |
Cameras for Node #1 | IP Range | Description |
---|---|---|
Camera 1 | 192.168.17.21 | Local Camera used within the 1st location |
Camera 2 | 192.168.17.22 | Local Camera used within the 1st location |
Camera 3 | 192.168.17.23 | Local Camera used within the 1st location |
Camera 4 | 192.168.17.24 | Local Camera used within the 1st location |
Cameras for Node #2 | IP Range | Description |
---|---|---|
Camera 1 | 192.168.17.31 | Local Camera used within the 2nd location |
Camera 2 | 192.168.17.32 | Local Cameras used within the 2nd location |
Camera 3 | 192.168.17.33 | Local Cameras used within the 2nd location |
Camera 4 | 192.168.17.34 | Local Cameras used within the 2nd location |
Cameras for Node #X | IP Range | Description |
---|---|---|
Camera 1 | 192.168.17.X1 | Local Camera used within the Xst location |
Camera 2 | 192.168.17.X2 | Local Cameras used within the Xst location |
Camera 3 | 192.168.17.X3 | Local Cameras used within the Xst location |
Camera 4 | 192.168.17.X4 | Local Cameras used within the Xst location |
We have more information available about setting up the IP addresses for Nodes and Cameras.
Node Installation
There are several requirements on an operation system and other software that are needed for successful installation of the SmartFace Platform on Linux.
SmartFace version 4.10 and higher versions require the CPU support AVX2 instruction set , eg Intel Haswell microarchitecture or AMD Zen family To have the Docker, Docker Compose and Docker CLI installed on your machine, Docker version 20.10.10 or higher and Docker Compose version 1.29.0 or higher. Generally, we recommend using the latest versions of the aforementioned software.
An operating system supporting above requirements is needed. A current Long Term Support (LTS) Ubuntu version would be recommended.
Once the operating system is installed, please follow the instructions on this link to install and run the Docker and Docker Compose.
Node Network Configuration
To set up the IP Address, Subnet mask and the Gateway, you can follow this guide.
SmartFace Installation
We need to install SmartFace on each node. For each node we need a license.
On your Nodes open the Terminal. Navigate to location /srv/
using the command:cd /srv
Then clone the git repositorygit clone https://github.com/innovatrics/smartface.git
Apply License
Each instance of the SmartFace needs a license. For more context and information about how to get and apply the license visit our Get Started documentation. Video guide is available.
Copy the license files into the folder central-node
for the central node:/srv/smartface/sf-docker/special/package-event-security-monitoring/central-node
or
Copy the license files into the folder local-node
for the local node:/srv/smartface/sf-docker/special/package-event-security-monitoring/local-node
Run the SmartFace
Depending whether this is a local or central node, run the run.sh
script located in the folder either in
/srv/smartface/sf-docker/special/package-event-security-monitoring/central-node
or
/srv/smartface/sf-docker/special/package-event-security-monitoring/local-node
For more context and infomation about installing read our Get Started documentation.
Once the installation successfully went through, the SmartFace is running. Visit the installation’s SmartFace Station using your own IP address, or localhost on the URL such as http://localhost:8000/
or http://your-ip-address:8000/
. Endure each node is up and running before proceeding.
Camera Installation
For the Event Security and Monitoring solution we will use the Edge Cameras equipped with Embedded Face Recognition, specifically using the SmartFace Embedded Stream Processor. They allows us to keep the server hardware requirements at the minimum, while providing high quality face recognition.
Power
Connect your camera into the POE enabled switch using a common ethernet cable. The recommended switches for local nodes do have POE enabled.
Camera Network Configuration
Set up the network settings of the camera to match your desired network architecture. Look at sample of network configuration. Freshly unpacked cameras do have a default IP Address set. Hanwha cameras have default IP set to 192.168.1.100
.
To change it, you need to connect the powered camera (POE) to a network, so it is reachable by your device, such as laptop, or one of the NUCs. If your network is not DHCP enabled, we suggest to set temporarily your device’s IP Address to be within the same range, such as 192.168.1.150
.
Visit the camera’s ip address, using it as URL in your browser http://192.168.1.100
, where you are prompted for Administrator password change. Set new password for the admin user as per password quality requirements.
Within the camera web UI, navigate to the Settings, Basic, IP & Port, where you can see the IPv4 and IPv6 setup. Use the IPv4, where you can define IP type (set to Manual), IP address, Subnet mask and Gateway to match the IP Address Configuration. Click Apply.
After succesfully applying the updated network configuration, you will need to connect to the new IP Address. To do so you will need to update the IP Address of the connected device (NUC/Laptop) to be within the right range. For this sample within the range of 192.168.17.*
.
Position
We are aiming at positioning your Edge cameras correctly to avoid common facial biometry challenges. More information about where and how to mount and setup a camera is available here. This includes information about how to set up focus, lighting, colors and zoom.
SFE Stream Processor
An Edge camera is a Smart camera using the SmartFace Embedded (SFE) Stream Processor. The SmartFace Embedded (SFE) Stream Processor is a face recognition pipeline processing camera input and performing real-time face recognition on edge devices, like smart cameras and AI boxes.
We will install the SFE Stream Processor on each Smart Camera. Please follow instructions below how to ensure the camera’s firmware is up to date and how to install the plugin.
Firmware
To successfully install the SFE Stream Processor Plugin for Hanwha, ensure you have the camera updated to firmware at least with the version: 2.21.06
and the Open platform has at least version 5.0
.
Install
Instructions on how to install SFE Stream Processor on supported Hanwha camera are available here. If you prefer a video guide, it is available below:
Installing SFE Stream Processor on a camera includes obtaining and applying a license for each camera. More information about the process of obtaining and applying licenses is available here.
Connect to MQTT Broker
Once the plugin is installed you can configure it to connect to the SmartFace MQTT Broker. The SmartFace MQTT Broker has the same IP Address as the SmartFace Platform server. Then click the LAUNCH button to start the SFE Stream Processor.
Populating Cameras into the SmartFace
Each installed camera needs to be added into the Local SmartFace Node. Please follow the steps to do so.
Configuration
Camera Configuration
Added Edge cameras can be configured. Detailed manual is available here. There are several recommended configurations optimized for the Event security. Please consider applying the configuration below:
- Face processing : Enabled
- Face save strategy: Balanced
- Min face size: 20px, Max face size: 300px
- Messaging : Enabled
- Strategy: On new and Interval
- Interval: Optimal (250ms)
- Send crop images : Enabled
- Send full frame images : Enabled
Define and set up watchlists
You can define your own Watchlists and enroll your own WatchlistMembers. This can be done using the SmartFace Station available at the port 8000
for each node, the manual is here or via the REST API. For more context, read about Watchlist Management and WatchlistMember Management.
Keep in mind the Watchlists can have their own colors to be easily distinguished and they can have their own biometric Matching Threshold. For a restricted Watchlists, such as “Blacklists” you can set the Watchlist color to be RED.
For Event Security Monitoring it is reasonable to set the Threshold to a lower value, for example 40 to be able to identify the most of the people walking in front of the cameras. A lower threshold might cause some false matches, however a possibility of catching someone from the “restricted” watchlist generally outweights the negatives or a general false match.
Choose data retention and saving strategy
To adjust the time for which data is retained, you can do the Data retention configuration for each node. The amount of days for storing data will limit the ability to investigate using the History Event. To avoid issues with using all the space on the server machine it is recommended to set a value. The maximum value might be influenced by your local laws. In general a good starting value would be 14
days.
How to Monitor Event Security
Add Preview cameras
When using the Edge Cameras, the notifications of events, such as face detection, match notification etc. are visible within the SmartFace Station. The Edge Cameras due to their nature do not produce and provide preview video. Such an enhanced preview video is possible using the RTSP cameras.
Smart Cameras using the SmartFace Embedded Stream Processor do also provide the RTSP stream that can be used by the SmartFace Platform and it can be visible within the SmartFace Station. Please keep in mind that encoding such video is consuming a lot of resources.
The recommened hardware is suitable for up to 2 preview cameras that can be used and previewed in the SmartFace Station. Please ensure that only cameras you are observing actually generates the Enhanced Preview.
To add RTSP streams for preview for each node you take a look at the instructions. You can enable the preview in the Camera Preview Settings
Notifications
In the SmartFace Station, the recent detections and identifications of individuals are showcased within the Recent Events section of the Security Dashboard on each Node for the added cameras. Each event notification corresponds to a single detected face, identified person, detected pedestrian or detected common object. If identification is successful, the event is presented alongside the stored image of the recognized watchlist member in the watchlist.
You can display the recent events for all cameras or only for selected ones. You can change this in the recent events settings.
For more information about Recent Events and Notifications you can look at the documentation
History Events
In accordance with your system configuration, the SmartFace has the capability to store data pertaining to detected faces, identified persons, detected pedestrian and other detected objects events. Within the history event page, you can engage in advanced filtering and searching functionalities across all the stored events, offering a comprehensive and tailored approach to accessing historical data.
Detailed information about the features and how to use them take a look at the documentation.