Our commitments
Our commitments: ethics is at the heart of our name
Spot relevant situations, protect data and keep people in control: these principles guide Perception’s design and the scope of every project.
VID · ÉT · ICS
Video · Ethics · Analytics
What technology can do. What it should rule out.
AI connected to a camera can count passages, detect a person on the ground or flag a growing crowd. Technical feasibility is only part of the question. We also need to establish what is necessary, proportionate and authorised. That is where we place ethics: between an image and how it is used.
On this page
1. Situations, never identities
4. Establish the framework before deployment
1. Situations, never identities
To help someone on the ground, an operator needs to understand the situation. The software does not need to establish who that person is.
Perception does not use facial recognition.
Analytics focus on observable objects, categories and situations.
Face localisation may be used for blurring, not for recognising identity.
Video search uses visible criteria such as clothing colour or vehicle type.
2. Images under your control
“Where do my images go?” is a fundamental question. The architecture must answer it for each feed, storage location and access path.
Local processing brings analytics close to your cameras.
With an on-site deployment, images are analysed on your infrastructure.
Exports, remote access and data destinations are agreed with you.
Retention periods, permissions and blurring options are defined for the use case.
Any use of images for research or training must be clarified in the project terms.
3. AI flags. People decide.
An alert draws attention; it does not replace an operator’s judgement. Context, verification and the decision on how to respond remain essential.
Analytics support human decision-making.
Operators check the evidence associated with an event before deciding how to respond.
The organisation defines qualification and escalation procedures.
Thresholds and scenarios are adjusted to reduce irrelevant alerts.
4. Establish the framework before deployment
The same feature cannot be used in the same way on a street, in a warehouse and at a station. Location, purpose and the people affected determine the framework to assess.
Technical capability is not legal authorisation.
Purposes, legal basis and required authorisations are assessed with the controller and its DPO.
Impact assessment, public information and individual rights are addressed under the applicable obligations.
GDPR, the AI Act and sector-specific rules must be assessed for the deployment in question.
5. Explain what AI does. And its limits.
Poor camera positioning, occlusion or a dark image: performance also depends on real-world conditions. A universal promise cannot replace an evaluation on representative scenes.
Performance must be checked against your use case.
Success criteria and evaluation conditions are agreed with you.
View limitations and unsuitable scenarios are identified.
Logging scope and available documentation are specified for your configuration.
6. Grow without losing our roots
Videtics has been part of the Technis group since 2025. Bringing expertise together does not remove the need to describe where processing happens and who can access data.
Perception continues to be developed in Sophia Antipolis, France.
Processing location, hosting and access are documented in the architecture.
Multi-site projects specify which data is centralised and where.
Network dependencies, maintenance and isolated-network operation are assessed during the technical study.
Our red lines.
Our approach to video analytics excludes the following uses.
Recognising a person by their face.
Inferring the origin, religion, opinions or sexual orientation of people on camera.
Claiming to read emotions or predict an offence from appearance.
Identifying participants in a demonstration from video images.
Repurposing data beyond the purposes agreed for the project.
Deploying analytics without assessing the necessary authorisations and safeguards.
Commitments you can examine.
These are the elements to review with your technical team, DPO and Videtics. Their content and availability are specified for your project.
Purposes and models
Enabled analytics, detected objects and processing objectives.
Architecture and network
Processing and storage locations, outbound flows and remote access arrangements.
Human oversight
Alert verification procedures and operator responsibilities.
Processing framework
Authorisations, impact assessment and information arrangements appropriate to the context.
Evaluation and records
Pilot criteria, results and the scope of available logs.
Hosting and dependencies
Data locations, services used and maintenance arrangements.
Prepare your documentation with your DPO
Contact us to identify the materials relevant to your project: processing descriptions, an architecture diagram, a network-flow matrix and information needed for your impact assessment. The documentation package is agreed with your team.
Questions about ethics and data
Does Perception use facial recognition?
No. Situation detection and searches using visible criteria are not intended to establish a person’s identity.
Does blurring make all data anonymous?
Not automatically. Identification risk depends on images, metadata and possible links with other information. It must be assessed in the context of the processing.
Is video analytics permitted everywhere?
No. The rules depend on the country, location and purpose. A feature being available in software does not authorise its use.
Does the software guarantee compliance?
No. Compliance also depends on the organisation, purposes and deployment. It is assessed with the controller and appropriate specialists.
Are my images used to train your models?
Purposes and permitted uses must be clear before any sharing. Request the terms applicable to your project, including trials or research work.
How do I obtain technical information for my DPO?
Use the documentation request link and describe your context. The Videtics team will direct you to the materials available for your configuration.
Video. Ethics. Analytics. Ethics stays at the centre.
Last updated : 15/10/2026 · Videtics, a Technis company.