How edge processing keeps your safety footage on-site
- 7 hours ago
- 4 min read
By Robert Thomson, Engineering Lead
One of the first questions I get from IT and security teams is a fair one. If you're putting AI on our cameras, where does our footage actually go? It's the right thing to ask. Camera footage is some of the most sensitive data a business holds, and people expect it to be handled with care.
The short answer is that with inviol, almost all of it stays exactly where it is. The footage is analysed on your site, by hardware that sits on your own network, and most of of your data never leaves the building. The technical reason that is possible comes down to one idea: edge processing.
What is edge processing?
Edge processing, sometimes called edge computing, means doing the computing work close to where the data is created, rather than shipping it off to a distant data centre first. Instead of streaming every frame of video to the cloud and waiting for an answer to come back, an edge device analyses the video on the spot and only passes the small amount of information that actually matters.
For most applications, that is a performance decision. For safety footage, it's a good deal more than that.
Why does edge processing matter for safety footage?
Stronger processing reliability. A safety system that stalls every time the internet hiccups is not a safety system you can lean on. Processing locally removes that dependency.
Less data leaves, so there's less to protect. A single camera produces an enormous amount of video. Sending all of it to the cloud, continuously, from every site, is expensive and it creates risk. Edge processing lets us filter on-site and send on only what's needed. In our case, that's a short clip and some event detail, not a 24/7 stream.
Cameras are a genuine security target. This is the part I care the most about as an engineer. Internet-connected cameras are an actively exploited target. The US cyber agency's catalogue of vulnerabilities under real-world attack includes camera flaws that let an attacker take over the device and use it as a foothold into the wider network. Streaming camera feeds to the cloud widens that exposure. Keeping the analysis on-site, behind your own network controls, keeps the most sensitive data out of the firing line. It is also why we treat data privacy on edge devices as an architecture problem, not an afterthought.
Where data lives is a compliance question. In New Zealand, the Privacy Commissioner is clear that CCTV footage is personal information, and advises that you collect the least amount of personal information that you can. In Australia, the Information Commissioner says recorded footage must be kept secure and destroyed or de-identified once it is no longer needed. Processing on-site, rather than shipping footage offshore by default, makes both of those obligations much easier to meet, especially if you operate across more than one country.
How inviol uses edge processing
The setup is deliberately simple. inviol connects to your existing CCTV cameras over standard feeds. A small hardware unit, leased as part of your agreement, sits alongside your recorder and runs the AI on-site, in real time. There are no new cameras to install and no sensors to bolt onto forklifts.
When the system spots a safety event, it does three things. It identifies the interaction that matters (a person, a vehicle, a zone, a speed), it blurs faces and people so the focus stays on the behaviour rather than the individual, and it saves a short clip with a risk rating and some context. That clip flows into a coaching queue for your team to review. The rest of the footage stays on-site and is not sent anywhere.
The 1% of data that does leave your site is held in an externally audited environment that meets SOC 2 and GDPR requirements. Read more on our blog post about protecting worker identities.
What does this mean for you?
For an operations or IT lead, the practical upshot is a shorter, calmer approval process. The privacy impact is smaller because most data never moves. The security review is simpler because you're not opening a continuous video pipe to the cloud. The Australian regulator's guidance on workplace monitoring and the NZ Privacy Commissioner's position both lean the same way: collect less, keep it secure, be transparent. On-site by default supports all three.
That's the quiet advantage of edge processing. It's not the feature anyone sets out to buy a safety platform for, but it's the reason the rest of the platform can be trusted with something as sensitive as footage of your people at work. If you want to see exactly what stays on-site and what does not, book a demo and we'll walk through the architecture with your IT and privacy people in the room.
Author
Robert Thomson
Engineering Lead
Frequently Asked Questions
Does my CCTV footage get uploaded to the cloud?
inviol analyses your footage on-site, in real time, on a hardware unit connected to your own network. Around most of your data never leaves the building. Only short clips are sent on for coaching to review.
What is the difference between edge processing and cloud processing?
Cloud processing sends data to a remote data centre to be analysed. Edge processing does the analysis locally, near where the data is created. For safety footage, local processing means faster detection, less data leaving site, and a smaller security and privacy footprint.
Can inviol access our footage?
inviol can access the event clips that leave your site for coaching. The clips that do leave are minimised, have faces blurred and are stored in a SOC 2, and GDPR audited environment. We think being clear about this matters more than a marketing claim that everything is untouchable.
Does on-site processing keep working if our internet goes down?
Yes. Because detection happens locally, the system keeps spotting safety events even if the connection drops. Even clips sync once the connection returns.
Is edge processing better for privacy compliance?
It helps. Privacy regulators in New Zealand and Australia both emphasise collecting the least data necessary and keeping it secure. Keeping footage on-site by default, rather than streaming it offshore, makes those obligations easier to meet, particularly across multiple jurisdictions.