Konect
Contact

Expertise

Professional video surveillance

A camera is worth nothing until someone has decided what its footage is for.

Parent division
Physical and cyber security

The problem

Most systems are sized in cameras rather than in uses.

A video surveillance tender almost always opens with a number. Thirty cameras, sixty, a hundred and twenty. That number is the outcome of a calculation nobody performed: it comes from an earlier quotation, an annotated site plan, or the available budget.

The useful question lies elsewhere. At every point of the site, it must be decided whether the image has to support detecting a presence, recognising someone already known or identifying an unknown face. These three uses call for different pixel densities on target, different optics, different camera positions and different lighting.

A system designed for detection and used for identification produces unusable footage at the moment it matters. The defect does not appear at acceptance: it appears on the day of the incident, when the recording is retrieved and proves nothing.

From intended use to evidence, schematic

01Intended use by zone
Detect, recognise or identify. The decision precedes every equipment choice.
02Camera and optics
Pixel density on target follows from the intended use, never from a catalogue.
03Segmented transport
A separate or partitioned network, sized on the real bitrate of the streams.
04Operation
Viewing station, per-user rights, extraction procedure.
05Recording
Volume calculated on the retention period required and the real bitrate, not a theoretical one.
06Extraction and audit trail
A sequence handed over in a form that stands, and a trace of every viewing.
Schematic. No camera count, no equipment and no real site appears on it.Typical sizing figures. Pixel densities on target follow EN 62676-4; the retention period is a common one, to be settled with the data controller.

Scope

What the service covers

Coverage study
Site survey, intended use defined zone by zone, pixel density on target calculated, camera positions and optics selected, a coverage plan that can be held against us at acceptance.
Capture infrastructure
Fixed and steerable cameras, power over the network, mounts and connections, mechanical protection, and treatment of viewpoints exposed to weather and backlight.
Transport and recording
Dedicated or segmented network, bandwidth and storage sized from the retention period actually required, redundant recorders, configuration backup.
Operation
Viewing station, video walls, per-user access rights, procedures for extracting and handing over a sequence, and an audit trail of every viewing.
Video analytics
Line crossing, abandoned object and loitering detection, tuned on site and observed over several day-night cycles before being brought into service.
Compliance and notice
Record of processing, signage informing people, documented retention period, and a procedure for responding to access and deletion requests.

Situations

The most frequent situations

Sensitive site with an extended perimeter
Fence line, vehicle access, technical areas. Perimeter coverage serves detection; forced passage points serve identification. Two regimes, one system.
Building open to the public
Halls, counters, circulation, emergency exits. The dominant constraint is informing people and confining fields of view to legitimate spaces.
Remote industrial or mining site
Intermittent lighting, dust, vibration, limited return link. Recording stays local and only requested sequences travel back.
Branch network
Identical installations replicated, supervision centralised, extraction performed remotely. The value lies in uniformity: a branch configured differently is a branch nobody knows how to operate.

Requirements

What to require, of us as of anyone

These requirements hold whichever supplier is appointed. Written into a tender, they filter out the responses that will not hold.

  • A coverage plan tying every camera to a use — detection, recognition or identification — rather than a list of part numbers.
  • The pixel density on target adopted for each zone, and the distance at which it is held.
  • The storage calculation, with retention period, compression ratio and the system’s behaviour once the disk is full.
  • Night-time performance measured under the site’s real lighting, on site, not copied from a datasheet.
  • The access rights regime for footage, and the audit trail of every viewing.
  • The procedure for extracting a sequence, the format it is handed over in, and who is authorised to produce it.
  • The operating documentation and the training that makes the system maintainable by another team.

Pitfalls

Common mistakes, and what they cost

Counting cameras before deciding uses
The number is fixed first and the uses adapt to it. The usual result is uniform coverage everywhere: too little where identification is required, more than needed elsewhere.
Sizing storage from the quoted compression ratio
The quoted ratio assumes a quiet scene. A busy scene produces several times the bitrate, and the real retention period falls below the contractual commitment without anyone noticing.
Treating video surveillance as a standalone package
Cameras consume bandwidth, backed-up power and switch ports. Installed after the network and the power supply, they force both to be reworked.
Leaving compliance to commissioning
Signage, register and retention period are decided with the architecture. Added afterwards, they sometimes require cameras already mounted to be moved.

Questions

Questions asked before consulting

How many cameras does a site need?

The question cannot be answered as a number. It is answered zone by zone, by deciding whether the image must support detection, recognition or identification. An identification zone demands a far higher pixel density on target than a detection zone, and therefore more cameras or different optics. The number is the result of that study, never its starting point.

What retention period should be adopted?

It follows from how long it typically takes for an incident to be noticed and reported, and from the rules applying to the site. A period chosen without that reasoning is either too short to be useful or too long to be justified. It drives the storage volume directly, and must be contracted against the real bitrate of the streams, not a theoretical one.

Do cameras need a separate network?

Separation is required, but it can be logical rather than physical. What matters is that video traffic does not degrade other uses, that cameras are not reachable from the office network, and that compromising one camera opens no path into the rest of the system.

Does video analytics replace an operator?

No. It reduces the volume to be watched by flagging events, which helps when the number of streams exceeds what an operator can follow. It produces false alarms, at a rate that depends on tuning and on the scene. Analytics commissioned without on-site tuning and without several day-night cycles of observation is abandoned by operators within weeks.

What should be planned for power?

Video surveillance loses its purpose at the very moment the power fails. Cameras, the switches that power them, the recording and the viewing station must be backed up together; backing up cameras without backing up the recorder retains no footage at all.

Can an existing system be upgraded without replacing everything?

Often. A coverage audit shows which cameras serve their intended use and which do not. The upgrade then touches a few viewpoints, the recording or the network, rather than the whole. A full replacement is justified when the cabling or the power supply can no longer carry the intended system.

Evidence

Where we have applied it

All projects

Neighbouring subjects

Solutions concerned

Consultations · Pre-qualifications · Partnerships

Let us discuss the actual case.

Describe the site, the dominant constraint and the deadline. We will say what requires a preliminary study and what can be committed directly.

Speak to an expertAll areas of expertise