Home › Comparison

Edge vs cloud

Edge or cloud video analytics: what actually changes.

Both run the same kinds of model. The difference is where the video is processed, and that one choice decides privacy, flexibility, scale, uptime and price.

What is the difference between edge and cloud video analytics?

Cloud video analytics uploads camera footage to a remote data centre to be processed there. Edge video analytics processes the footage on a device or server inside the building and sends out only the result. Edge is faster because nothing waits on an upload, it keeps working without internet, it costs the same whether a site has five cameras or fifty, and the video never leaves the premises.

DimensionCloud video analyticsTarsyer edge
Where video is processedUploaded to a third-party data centreOn a device or server inside your premises
Time from event to alertSeconds to minutes, after upload and queueingUnder a minute, usually seconds
When the link dropsDetection stopsDetection continues, alerts queue
Bandwidth per camera1 to 4 Mbps, continuouslyKilobytes per alert
Going from 10 sites to 1,000Central processing, storage and bandwidth all growEach site brings its own processing. Nothing central changes
If hardware failsA data-centre outage stops every site at onceAnother device takes over at sites with more than one. Other sites are unaffected
Cost as cameras are addedGrows with streaming and storageFlat, one device covers many cameras
Who holds the footageThe vendorYou, because nothing leaves the site
Typical IT reviewWeeks, and often blockedShort, because there is nothing to review
New cameras requiredOften, for cloud-native hardwareNone, existing CCTV is used
Hardware on siteNone, but every camera streams out continuouslyOne small device, or an entry-level workstation. No graphics card

Where cloud wins, honestly: a single site with a handful of cameras and no privacy constraint is simpler in the cloud, because there is no hardware to install at all.

Beyond privacy

Why chains choose edge.

Keeping video on site is the reason IT signs off. These are the reasons operations teams stay.

Privacy

Video stays in the building.

Frames are processed on site and discarded. Only alerts, snapshots and counts leave, so there is nothing for IT to review.

Flexibility

Each site runs what it needs.

Queue alerts in one store, hygiene checks in another, PPE on one line only. Checks are set per site and per camera.

Scale

10 stores or 1,000, nothing changes.

Every site brings its own processing, so there is no central server to upgrade and the bandwidth bill stays flat.

Redundancy

One device fails, the rest carry on.

At sites with more than one device, its cameras move to another and the checks keep running. A fault in one store never touches the rest.

The architecture

Video stays in the building. Only the answer leaves.

The Tarsyer device reads your recorder, or the cameras directly, over the local network, runs the checks, and discards the frames. What leaves the site is an alert, a snapshot and a count.

YOUR PREMISES Camera 1Camera 2Camera n Existing NVR / DVR Tarsyer edge device runs all checks Video is processed here and discarded. Nothing is uploaded, nothing faces the internet. alerts & counts WhatsApp / emailDashboardHooter / PA Snapshots and numbers, kilobytes per alert.
Questions

Edge and cloud, answered.

Edge is better wherever the video is sensitive, the connection is unreliable, or the camera count is large. Processing on site means detection in seconds rather than after an upload, it keeps working when the link drops, and bandwidth stays flat as cameras are added. Cloud makes sense when a site has very few cameras and no privacy constraint.

Streaming one camera continuously to the cloud costs roughly 1 to 4 Mbps depending on quality, and that multiplies by every camera. Edge processing sends only alerts, snapshots and counts, which is kilobytes per event, so a hundred sites cost about the same in bandwidth as one.

Yes. Detection runs on the device inside the building, so it continues through an outage. Alerts queue and are delivered when the connection returns. Internet is needed to deliver a message, not to spot a problem.

No. Edge means the video is processed on your premises instead of in a data centre. Tarsyer usually supplies a small edge device for each site, but if you already have a server or workstation on site, Tarsyer can run on that instead. It needs no graphics card, and either way the video never leaves the building.

Yes, and more simply than cloud. Each site brings its own processing, so going from 10 stores to 1,000 needs no central server upgrade and no bigger internet link, and each store can run a different mix of checks. Devices update themselves in staggered waves, so a larger rollout does not change how the chain is run.

At sites with more than one device, another device takes over its cameras and the checks keep running. Either way, the failed device reports the fault so it can be replaced, and because every site processes its own video, a failure in one store never affects another, unlike a data-centre outage that stops every site at once.

Materially, because no video leaves the premises. There is no footage sitting with a third party to review, which is usually what shortens the IT and compliance conversation from weeks to a call.

See it running on your own cameras.

A 30-minute call. Bring a floor plan or a camera list and we will tell you which checks fit, what it costs and how soon it goes live.

Or write to sales@tarsyer.com