CCTV & Surveillance in Guwahati.

Every camera position answers one of four questions — the DORI scale in IEC/EN 62676-4, the standard that turns “good coverage” into a number a tender can be judged against. Detection asks whether something is happening. Observation asks what kind of activity it is. Recognition asks whether this is a person known to the site. Identification asks who this is, to a standard that would survive being questioned. Those four intents demand progressively more pixels across the target, and a camera schedule that does not state the intent per position is a schedule nobody can price or verify.

That intent, the working distance and the lens together determine pixel density on target — DORI states it in pixels per metre of scene width, and it is pixel density, not headline sensor resolution, that decides whether footage answers the question. A high-resolution camera covering a wide forecourt can deliver less usable detail at the gate than a modest camera aimed properly at the gate. This is the single most common gap between what a client expected and what the recording actually shows.

The second axis is the retention and recording decision. Continuous recording, motion-triggered recording and event-driven recording produce very different storage profiles and very different evidential value, and the retention period should come from your own policy or from what a plausible investigation would need — not from whatever the storage happens to allow.

Surveillance footage is personal data, and the design has to reflect that. Areas of legitimate privacy expectation are excluded by camera placement or by privacy masking at the encoder; who may view live and recorded footage, who may export it and under what authority, and how long recordings are kept before automatic deletion are all policy decisions that belong in the design brief rather than in a later argument.

Evidential integrity is the technical counterpart. Time synchronisation across every recording device, an audit trail on export, and a defined chain-of-custody procedure are what make a recording useful to somebody who was not there. A system where cameras disagree about the time produces footage whose sequence cannot be established.

Signage and staff notification are part of a lawful operating posture, not decoration. The system design, the written policy and what people are told about it need to agree — and where a site operates under a client's own security policy, the design is subordinate to that policy and should be reviewed against it before installation rather than after.

Sightlines are an architectural conversation and they close early. Columns, signage, planting that will mature, roller-shutter housings and the swing of a door all obstruct a camera that looked correct on plan, and the time to resolve that is while the plan is still moving. Mounting height is the same conversation: too high and faces are foreshortened past usefulness, too low and the camera is reachable.

Lighting is the second and it is a genuinely cross-disciplinary item. A camera facing a glazed entrance with daylight behind it is contending with backlight the lighting designer never considered; a car park lit to a uniform low level may be adequate for people and inadequate for recognition. Where the lighting scheme cannot serve the surveillance intent, supplementary illumination becomes part of the design and needs its own power and control.

Power and network provision is the third. Powered-over-Ethernet cameras need a switch budget that accounts for every device's actual class and for the heaters and blowers on external housings, plus headroom; the recording position needs power, cooling and physical security. All of that is a containment and small-power decision taken with the electrical contractor at first-fix, not a commissioning discovery.

Monsoon rainfall defeats external installations that were specified for a milder assumption. Water tracks along cable into the housing through an unsealed gland, condensation forms inside a dome that was opened on a humid afternoon and never dried, and the resulting internal fogging is invisible on a commissioning check made the same week and obvious three months later. Enclosure ratings, gland discipline and the sealing procedure at installation matter more here than the camera specification does.

Heavy rain also degrades the image itself. A camera relying on infrared illumination in driving rain illuminates the rain, and the resulting scene is unusable exactly when it is most likely to be needed. Positions that must work through the monsoon need that condition designed for, not assumed away.

Power quality is the third mechanism. Voltage instability shortens the life of camera power supplies and network switches, and a recorder that loses supply mid-write can corrupt a recording segment. The recording position's power arrangement — conditioned supply and battery-backed continuity sized to ride through the interruptions this grid actually produces — is a design decision rather than an accessory.

Lightning and surge reaches cameras through long external cable runs and through inter-building links, and a single strike can take out a run of devices and the switch port they terminate on. Surge protection at building transitions is standard practice on a distributed site here.

Finally, the quiet failure: nobody checks the recordings. Discs fail, cameras drift out of focus, a repositioned sign obscures a view, and none of it announces itself. A system without a scheduled verification routine is discovered to be broken on the day somebody needs the footage.

Every camera verified against its stated intent at its stated distance, in the light conditions it will actually work in — including after dark, and where relevant during rain. A daytime-only commissioning check certifies half a system. Focus and back-focus set and recorded, so that a later drift is detectable rather than gradual.

Recording verified against the retention requirement: not that recording works today, but that the storage will still hold the required number of days under the activity the site actually generates. Time synchronisation confirmed across every device. User roles, permissions and the export workflow tested with the people who will use them.

Handover carries the as-built camera schedule, network and licence documentation, the operating and privacy policy, the verification routine with its cadence, and a signed training record. The AMC then states the health-check cadence, what is inspected physically versus remotely, the firmware-update position and the response arrangements — because the failure mode that matters most in this discipline is the one nobody notices until the footage is needed.