Server Based AI vs. Edge Based AI: what are the differences?

The popularity of AI within video analytics continues to grow. Dat komt doordat deze technologie toegankelijker wordt en steeds breder toepasbaar is. Video AI Analytics kan op verschillende manieren gehost worden: via een server of op het device zelf. Maar welke optie is voor dan het meest voordelig? In deze blog bekijken we wat de verschillen zijn en wat deze betekenen als je besluit te investeren in Video AI Analytics.

Before we delve deeper into the differences, it’s good to take a quick look at what the definitions of Server Based AI and Edge Based AI are.

What is Server Based AI?

Server Based AI actually kind of says what it stands for. With Server Based AI, the AI technology is managed centrally on a server, which can be either local or in the Cloud. Server Based AI is especially beneficial when there is a large amount of data to be processed. An example of this is when many cameras are connected or complex AI algorithms are used, resulting in a large amount of data being analyzed simultaneously. Server Based AI uses existing camera streams to apply AI technology to.

What is Edge Based AI?

In Edge Based AI, the AI technology is hosted on the equipment (the camera) itself. The equipment is thereby enriched with a chipset, which imparts intelligence to the camera. Analysis and detection therefore takes place directly on the camera.

Server Based AI vs. Edge Based AI

So from the definitions, it appears that both Server Based AI and Edge Based AI perform the same task, just in a different way. It’s good to know what this means before getting started with Video AI Analytics. To give a good overview, we list several points that are important below.

Applicability to the existing situation

The first major difference between Server Based AI and Edge Based AI, is in the applicability within the existing situation at your location. Because where Server Based AI uses existing camera streams, Edge Based AI analysis and detection takes place on the equipment itself. For the latter, however, the equipment must be suitable. If it is not, then the cameras must be replaced. With Server Based AI, on the other hand, it is only necessary to connect existing cameras to one central server. A lot more accessible, in other words.

ROI: one-time and long-term

That applicability also does something to your ROI. If you choose Edge Based AI, then your initial investment is also likely to be higher. Your current camera network will most likely need to be replaced. If this involves dozens or hundreds of cameras, this can be quite expensive. Connecting existing cameras to an AI server with associated software often requires a much lower investment.

In addition, the ROI over the longer term with Edge Based AI will also be higher. Because no matter how fast the developments of the various chipsets and smart cameras go, what is state-of-the-art now will be obsolete in three years. This means that the Edge Based AI features that will come to market in a few years will no longer be compatible with the current chipsets on the cameras. This makes replacement necessary. An AI server also needs to be replaced once in a while as well, but that investment remains lower than replacing the entire camera network.

The number of AI functions on one camera

Another difference between Server Based AI and Edge Based AI is the number of AI functions that can run on a single camera. In fact, with Edge Based AI, we often see that the chipsets that make the cameras smart can often only run one AI function at a time. That limits the possibilities. With Server Based AI, the situation is different. There, it is possible to deploy cameras for different purposes. For example: Intrusion Detection at nights, butCrowd Management, abnormal behaviour and falling people at daytime.

Phase 3 versus phase 4 AI

AI is evolving at a rapid pace. Detection is currently mainly based on people, objects or vehicles. In the near future, it will increasingly be based on patterns and behavior. Think of the detection of aggression or the optimal management of large crowds of people. This requires more and more computing power, which automatically makes Server Based AI ahead in this area. After all, greater capacity means earlier chances to adopt previous developments.

Independence

A key difference between Server Based AI and Edge Based AI, is independence. Edge Based AI is often developed by camera vendors themselves, which means that these Video AI features are only applicable to the cameras of this particular vendor. Server Based AI is basically camera- and brand-independent. Good to know, should your camera network consist of different brands and/or types. This can be an advantage if you, as a system integrator or installer, run multiple camera brands.

Cyber Security and Customizing

Server Based AI and Edge Based AI also require completely different approaches to each other when it comes to cyber security and protecting the metadata generated. Securing a single AI server requires a different approach than securing multiple networks of smart, stand-alone cameras.

Finally, Server Based AI also lends itself better to customization. For example, if you want to learn new objects, this is often possible with Server Based AI. This is often not the case with Edge Based AI. There, the various chipsets often come with non-customizable AI Analytics. This makes it just a little less flexible.

Conclusion

Even though Server Based AI and Edge Based AI are used for the same purposes, there are quite a few basic differences. Differences that should be carefully considered should you make the choice to apply AI for video analysis.

Want to know more about the above differences and what best suits your situation? Feel free to contact us. We would be happy to help you further. At VAIBS, we like to introduce companies to Video AI Analytics and the possibilities of our Vaidio platform in an approachable way. Want to know more about Vaidio now? Visit the the product page or download the brochure.