First things first

A scalable interface to manage all your agents and sites through a single pane of glass.

When installing a Kerberos agent inside your local network, it’s possible to review your activity by using the agent’s web interface. By nature, you will not be able to access the web interface from the cloud (outside your local network). On top of that, the Kerberos agent doesn’t come with a consolidated overview. This means that you will need to open multiple web interfaces, one for each connected surveillance camera.

To simplify and resolve all of these challenges, Kerberos Hub was developed.

Kerberos Hub is a cloud solution which allows you to have a single overview of all the media recordings generated by your surveillance cameras. It’s not a solution for processing your surveillance cameras in the cloud. For processing of your surveillance cameras, a Kerberos agent is used.

A brief list of the key features of Kerberos Hub are:

  • Secured storage
  • Consolidated overview
  • Advanced searching
  • Machine learning
  • Alerts
  • Live streaming
  • Accounts
  • Logging

Checkout the demo environment to see and feel the Kerberos Hub solution yourself.

Architecture

A simplified architecture is found below. In this architecture the Kerberos agents are installed on-premise, next to the surveillance cameras. Once one of the agents recorded something it will send the recording to our Kerberos Vault, using the credentials you’ve defined inside the agent.

As soon as the recording is succesfully received several engines will be executed, that will further process the recording and extract metadata from it (classification, origin of the surveillance camera, etc). A lot of processing is happening behind the scenes (by what we call microservices).

By using a browser you can navigate to Kerberos Hub, and review all the recordings made by the Kerberos agents. On top of that you will receive more context about the recordings, get notifications, view a livestream and much more.

Kerberos Hub is composed out of of different microservices.

Kerberos Hub is composed out of of different microservices.

Public or Private edition

They are two different versions of using and deploying Kerberos Hub, we have the public and private dition. What’s the difference?

The Public edition of Kerberos Hub, is running all its services in what we often call the public cloud; and to be more specific on AWS and GCP. The Public cloud is mainly built for people who want to have a low cost and small scale VMS in the cloud. It’s easy to (un)subscribe to the system, and you can connect your Kerberos agents in a matter of seconds.

On the other hand we also have the private edition, which is available for enterprises. In contrary to the public edition, a customer can deploy the entire Kerberos Hub platform (with all its microservices) where it wants (on-premise or cloud). Both public and private edition are built on Kubernetes, this means that you will need to have a Kubernetes cluster available at the location you which to install Kerberos Hub.

For more information about the licensing of Kerberos Hub private edition, please get in touch with cedric@verstraeten.io. The licensing model of Kerberos Hub public edition can be found online.

Features

So you might wonder, what’s actually inside the platform? Please find a concrete list of functionalities below, this should give you more insights about how it looks like and what the capabilities are.

Secured storage

Kerberos Hub allows you to store your data in the cloud, and make it available using a secured web interface. This is how it works.

Once you’ve installed a Kerberos agent, and created an account on Kerberos Hub, you will be able to connect both systems together. By entering credentials received from Kerberos Hub, into your Kerberos agent, your media recordings will be send to our cloud storage.

You can link one or more Kerberos agents to your Kerberos Hub account.

After linking your media recorded by a Kerberos agent will be send to our secure cloud storage, Kerberos Vault. By default our own Kerberos Vault service is used, however you can install your own Kerberos Vault and link it to our public cloud. By doing this all the recordings are stored in your own datacenter of choice, and only metadata is stored inside Kerberos Hub.

You connect your agents to Kerberos Hub, through Kerberos Vault.

You connect your agents to Kerberos Hub, through Kerberos Vault.

Once stored your media recordings will be processed by different functions and algorithms to send alerts, apply machine learning and much more.

Kerberos Hub is much more than a storage service, you get intelligent capabilities as well. Read on to discover more.

Consolidated overview

Kerberos Hub will group your media recordings (coming from one or more Kerberos agents) into time bundles. We do this to simplify the searching of a specific event. Less data is shown at once, and when needed you can drill down a level deeper to see more details.

The media page gives you an overview of all your recordings.

The media page gives you an overview of all your recordings.

In a sneak peek you will see how many media recordings were generated during a certain time period, and which Kerberos agents were involved. For example, this might give you an indication of a particular event (more about in the machine learning section).

Security

Once you’re inside your Kerberos Hub account, you will see all the available media recordings of your Kerberos agents in a single interface. So how does this work?

When you open Kerberos Hub we will generate temporary and private links for every media recording within your account. This link includes a random and unique hash, and is only available for a couple of minutes. So this means that your recording cannot be shared afterwards, or the risk someone else accessing that link is minimised.

Advanced searching

Next to showing all your recordings in one overview, we also give you the capabilities to search for a recording more easily. For example you can search for recordings within a specific time boundary, recordings generated by a specific Kerberos agent, recordings which are marked as important, classified recordings such as pedestrians or animals, and more.

Detailed searching across different sites and cameras. Using smart search to look for specific objects.

Detailed searching across different sites and cameras. Using smart search to look for specific objects.

Machine learning

With Kerberos Hub you will get far more than just a nice interface for finding back your recordings. When a recording is uploaded to Kerberos Hub, we will execute all kind of algorithms. One of these algorithms is classification. This algorithm is used to classify/recognise all the objects inside your recording. Any human, vehicle or animal in the recording will be labeled.

Next to classifying, the algorithm is also tracing the classified objects. By doing this we can label the recording with the object of interest. For example at the image below you see the traject of a pedestrian walking by (710 pixels).

When running your recordings on our Kerberos Hub, it will be processed by our ML service.

When running your recordings on our Kerberos Hub, it will be processed by our ML service.

You can filter and drill down into specific media where a classification was detected.

You can filter and drill down into specific media where a classification was detected.

Alerts

When a specific event occurred, a notification can be send to one or more channels. A user can compose an event by selecting one or more conditions. Once the event is triggered, a notification is sent to all the enabled channels.

Alerts

Kerberos Hub supports different types of alerts:

  • Detections: send a notification if a recording was received from a specific Kerberos agent, within a time range, with a specific classification, etc.
  • Devices: when one of your Kerberos agents stopped working, a notification can be send.
  • High upload: when a lot of recordings are generated within a specific period of time, a notification can be send.
Configure specific scenarios to trigger alerts.

Configure specific scenarios to trigger alerts.

Channels

Kerberos Hub supports different types of channels, towards notifications can be send:

  • Browser: notifications inside the browser (if the used browser supports this).
  • E-mail: a personal email can be send.
  • Slack: send a message to a slack channel.
  • Alexa: send a notification to Alexa skill (Notify Me).
  • Telegram: send a message to a Telegram channel.
  • Pushbullet: send a message to a Pushbullet channel.
  • Pushover: send a message to a Pushover channel.
  • Webhook: trigger a webhook (POST).
  • IFTTT: send a message to IFTTT.
  • SMS: send a text message to a mobile number.
Channels allow you to communicate to external services.

Channels allow you to communicate to external services.

Livestreaming

Kerberos Hub allows you to live stream from your Kerberos agents to your Kerberos Hub account.

For Kerberos Open Source a low resolution livestream is made available, this will allow you to see what is happening in front of your surveillance camera in real-time. The stream is shared over MQTT (TCPS) and visualised in the Kerberos Hub webinterface using secure websockets (WSS). No port forwarding needs to be enabled.

A Kerberos agent will give you access to the livestream, and a HD stream when using the enterprise agent.

A Kerberos agent will give you access to the livestream, and a HD stream when using the enterprise agent.

For Kerberos Enterprise a HD livestream is made available. The full resolution and FPS is send through a concept called WebRTC. This WebRTC technology, is used my all major conferencing tools such as Zoom, Teams, etc.

Accounts

It’s possible to create multiple accounts (also called subaccounts) within your main account. Their are two different roles:

  • Guest: will only see your media recordings. A guest cannot modify, delete or create something.
  • Editor: has full access, same permissions as the main account.
Using accounts you can give external people access.

Using accounts you can give external people access.

Edit this page on GitHub