Ayonix Face Recognition

Tell us about the deployment

Everything marked required is needed to scope a response. The camera count, gallery size, deployment preference and platform are asked because those four answers decide whether a pilot can be designed at all — not to qualify anyone out.

Analysed streams, not the total camera estate.

1:N error rates rise with gallery size, so this changes the design.

The environment, the people who pass through it, and the decision the system has to make. Two or three sentences is plenty.

We reply from engineering rather than from a queue. No accuracy percentage will be quoted, because there isn’t one that would be true for your site. If a sample image or clip would help, there is a separate secure upload below.

Optional: send a sample image or clip

A still or a short clip from the camera you want to use, taken at the time of day that matters, tells us more than any specification sheet. Thirty seconds is plenty.

Before you upload

A camera view contains images of real people who did not choose to send them to us. Faces are sensitive personal information. Only upload media you are authorised to share, and consider whether a view with fewer people in it would answer the same question.

  • Stored privately and never served back over the web.
  • Deleted automatically after 7 days, by a scheduled job.
  • Used for a technical assessment of the camera view, and nothing else.
  • Never used to train any model.
  • No recognition is run on it automatically.
  • Access is restricted and every access is logged.

JPEG, PNG, WebP, MP4 or MOV. Up to 25 MB.

This upload is optional. Everything on this page works without it, and describing the camera view in the enquiry form is a perfectly good alternative.

Frequently asked questions

What happens after I submit this?

Someone who can discuss capture geometry, threshold selection and integration reads it — not a queue. The first reply asks about the things that decide feasibility: what the camera sees at the point people pass, what the fallback path is, and which system has to receive the result.

Will I be given an accuracy figure?

No, and that is deliberate rather than evasive. This site publishes no accuracy percentage for any vendor including Ayonix, because a figure without its threshold, dataset, gallery size and demographic breakdown cannot be reproduced. What you will be offered instead is a measurement on your own cameras with acceptance criteria agreed in advance.

Do I have to upload a sample image or video?

No. The upload is optional and everything works without it. It is offered because a thirty-second clip from the actual camera answers questions that a specification sheet cannot — but describing the view in the form is a perfectly good alternative, and for some sites it is the only permissible one.

What happens to a sample I upload?

It is stored privately, never served back over the web, used only for a technical assessment of the camera view, never used to train any model, and deleted automatically within seven days by a scheduled job rather than by anyone remembering. Access is restricted and logged. You get a reference you can quote to have it deleted sooner.

What do you do with the information in this form?

It is stored so that we can answer your enquiry, and it is used for that. The camera count, gallery size, deployment preference and platform are asked for because those four answers determine whether a pilot can be scoped at all — not to qualify you out. The privacy notice sets out retention and your rights.

What if we are not ready to buy anything?

That is a normal and useful place to start a conversation. Most of the value in an early discussion is in establishing whether the cameras can support the requirement and what the fallback path costs to staff, and both are worth knowing before a budget exists.