Privacy by Design
The Camera That
Never Saw a Thing.
Most camera-based AI systems treat privacy as a policy — a promise made in a document, honoured until a breach, a misconfiguration, or a subpoena. Fluxor Vision treats privacy as an architecture. The guarantee holds regardless of what happens downstream.
That said, not every deployment should discard the image. Quality inspection needs the frame. Security forensics depends on it. Audit trails sometimes require visual evidence. We support both approaches — traditional image pipelines and on-sensor inference — and help you choose the architecture that fits your context, with a clear view of what each carries with it.
Two architectures, two different sets of trade-offs. Fluxor Vision supports both — the right choice depends on what your deployment actually needs.
Image retention makes sense here
Structured output is sufficient here
⚖️ Data Minimisation — Article 5(1)(c)
GDPR requires that personal data be adequate, relevant, and limited to what is necessary for the stated purpose. On-sensor inference is a direct technical implementation of this principle — if the purpose is to count occupants, producing a count satisfies the purpose. Producing a video archive does not.
🏗️ Hardware Guarantee vs Policy Promise
A data protection policy states what an organisation intends to do with data. A hardware guarantee defines what the system is architecturally incapable of doing. When pixel data is discarded on-sensor, no policy violation, misconfiguration, breach, or legal compulsion can produce footage that was never created.
📋 Is Inference Output Personal Data?
A count of persons in a room is unlikely to constitute personal data. A precise skeletal keypoint sequence associated with a named individual may. The output format and combination of signals determines exposure — not the hardware alone. Seek legal review for any inference model whose output could identify or profile individuals.
📝 Responsible Deployment — Both Approaches
Regardless of pipeline choice, responsible deployment includes: a Data Protection Impact Assessment scoped to the context; clearly defined retention windows for any stored data; transparency signage where individuals may be present; and a legal review of inference outputs against applicable data protection law.
Full Capability Map
15 Deployments.
Right-Sized for Each Context.
Every capability below is available across both pipeline approaches. The badge indicates the natural fit — but every deployment is a considered decision, not a default.
Responsible deployment — whichever approach you choose
Every Fluxor Vision engagement includes a configuration review against your stated use case, guidance on Data Protection Impact Assessments, retention policy recommendations, transparency signage requirements where individuals may be present, and a frank conversation about whether image capture is justified by the deployment context — or whether structured output is sufficient. We do not have a default answer. We have a process for arriving at the right one.