Home › Resources › Choosing a Body Camera
Nine decisions settle it, and only two of them are about picture quality. This guide works through them in the order that actually narrows the field, using the capability range we supply in the UAE, so you can arrive at a specification before you arrive at a model.
BodyCamera.ae Technical Team
Start here
Three questions eliminate most of the range before any specification is compared. Answer them first and the shortlist usually writes itself.
Written for security managers, facilities teams and procurement staff drafting a specification. Figures are the range we supply in the UAE, quoted as ceilings and floors. Which model fits is a short conversation once the requirement is settled.
Every other decision is downstream of this one. There are three architectures and they have different costs, different failure modes and different regulatory consequences.
The trap is buying cellular for a deployment that never needed it. A live feed is worth paying for when somebody is watching and can act on what they see. If no one is watching in real time, cellular adds a SIM, a data plan and a battery cost for a file that would have arrived at the end of the shift anyway.
The opposite trap is buying offline for a team that turns out to need escalation. Retrofitting connectivity means replacing the hardware, so the question deserves a real answer at specification time and not a hopeful one.
Whichever architecture you choose, the offload path should not have a manual step in it. A file that an officer can hold back, trim or delete before it lands is a file with a gap in its chain of custody, and that is the most common reason footage fails when it is finally needed.
Published runtimes across the range we supply run from nine hours to twelve, with one helmet-mounted unit reaching thirteen. Those figures are continuous recording at the model's standard resolution; recording at a higher resolution, streaming over cellular, or running infrared through a night shift will all pull the real number down.
The more useful distinction is how the battery is arranged, because it decides what happens when the shift outlasts the cell.
| Arrangement | What it means in the field | Suits |
|---|---|---|
| Fixed internal cell | Sealed into the device. Simplest, cheapest, one fewer thing to lose. When it is flat the shift's recording ends | Shifts that finish inside the rated runtime and end at a dock |
| Swappable cell | The officer carries a spare and changes it. Recording stops for the swap, so there is a short gap in the record | Long shifts where a brief gap is acceptable and can be logged |
| Hot-swap with a bridge cell | A small internal reserve keeps the camera running while the main cell is changed, so recording continues through the swap | Continuous-coverage posts and any deployment where a gap has to be explained |
Scroll the table sideways on a narrow screen.
For a standard eight-hour post with a dock at the end of it, a fixed battery is the right answer and the cheapest one. For a twelve-hour post, or a shift that can be extended without notice, the swap capability is what stops the record ending before the shift does.
Battery chemistry also degrades faster in UAE conditions than the datasheet assumes, because heat accelerates it. Budget for cell replacement inside the deployment's life, not at the end of it, and specify the spare count at order while the pricing is still open.
Across the range we supply, recording runs from 1280 x 720 up to 3840 x 2160. Higher is not automatically better, because resolution is the one specification that keeps costing money after the camera is bought.
Every step up multiplies against three things at once: the storage array, the time it takes to offload a shift, and the bandwidth if footage moves over a network. Quadrupling the pixel count roughly quadruples all three. The estimator below turns that into a number for your own fleet.
| Recording resolution | What it is good for | Relative storage per hour |
|---|---|---|
| 1280 x 720 | Confirming what happened and who was present, where the people are known | Lowest. Roughly half of 1080p |
| 1920 x 1080 | The working default. Recognising a face at conversational distance and reading close detail | The baseline everything else is measured against |
| 2560 x 1440 | Identifying an unfamiliar face at a few metres, and holding detail when footage is cropped | Roughly double 1080p |
| 3840 x 2160 | Reading plates and small text at distance, and evidence that will be enlarged | Roughly four times 1080p |
Scroll the table sideways on a narrow screen.
The codec matters as much as the resolution. H.265 delivers roughly the same picture in about half the data of H.264, so a fleet on H.265 at 1080p can cost less to store than a smaller fleet on H.264. Variable bitrate helps again, because a static scene stops consuming the same data as a moving one.
Internal storage across the range runs from 16 GB to 512 GB and on most models it is specified at order. That is worth settling early: capacity is usually fixed at manufacture, so it is not a decision you can revisit once the fleet is deployed.
Retention is the other multiplier, and it is usually set for you. Regulated premises in Dubai work to a 31-day floor with capacity computed above the calculated requirement, and high-risk sites to 90 days.5 Those figures are written for the fixed system installed at a premises, so treat them as the reference point your own retention policy has to justify itself against.
A practical way to choose: work out the smallest resolution at which the footage still answers the question you are buying it to answer. If the requirement is recognising a person you already know, 1080p is generous. If it is reading a plate or identifying a stranger at distance, the extra pixels earn their storage.
Lenses across the range publish horizontal fields of view from 103 degrees to 150, and one helmet-mounted unit uses a 90 degree rotatable head so the officer can aim it. Two models publish no figure at all, which is worth knowing before a tender asks for one.
A wide lens captures more of the context around an incident, which is usually the point. It also spreads the same pixels across a wider scene, so a face at ten metres carries fewer pixels at 150 degrees than at 110. And the wider the lens, the more visible the distortion at the edges, which matters if the footage will be used to judge distance or position.
For close-quarters work such as retail, reception and passenger contact, a wide lens is the right choice because the subject is near and the context is the evidence. For perimeter and open-area work, a narrower lens holds detail further out. Where an officer needs to aim at something specific, the rotatable head does a job no fixed lens does.
Most incidents that end up in front of a lawyer happened in poor light. The range solves it three ways and they are not interchangeable.
| Approach | How it behaves | Trade-off |
|---|---|---|
| Infrared illumination | Invisible lamps light the scene; the image switches to monochrome. Published useful range across the models that quote it runs to about ten metres | No colour, so clothing and vehicle colour are lost. The lamp is invisible to the subject, so it does not warn them |
| Starlight sensor | A more sensitive sensor holds usable colour in very low light without any lamp | Needs some ambient light. In true darkness it has nothing to amplify |
| White-light illumination | A visible lamp lights the scene and colour is retained | Announces the camera's presence and consumes battery quickly. Sometimes the point, sometimes a hazard |
| Dual infrared with a laser assist | Extends useful infrared range and helps the sensor focus in the dark | Available on part of the range only, and worth confirming per configuration |
Scroll the table sideways on a narrow screen.
On at least one unit in the range the choice between infrared and starlight is made at order and cannot be changed afterwards, so it belongs in the specification and not in the deployment. Where a team works both lit and unlit environments in the same shift, say so early; that constraint changes the shortlist.
Sensitivity down to around 0.2 lux is the benchmark the better sensors reach before the illuminator has to do the work. Below that, what you are seeing is the lamp, not the sensor, and the useful range of the lamp is the real specification.3
Chest mounting is the default across the range and it is the right default. It sits at roughly the height of the interaction, it is visible to the person being recorded, which is a legal requirement as well as a courtesy, and it survives being knocked.
Shoulder and epaulette mounting raises the view above a crowd and above the officer's own arms, which matters where hands and objects are what needs to be in frame. Helmet mounting follows the head rather than the torso, so it points where the officer is looking; that is the right answer for motorcycle patrols, height work and confined spaces, and the wrong one for a static post.
Whatever the mount, the fixing is what fails first. A camera that comes loose in the first month gets left in a drawer, and a camera in a drawer records nothing. Specify the mount against the actual uniform, and buy spares of the clip rather than of the camera.
The mount is also a compliance question. Rules for enforcement work require the camera to be in clear sight, so a concealed mount is not available in those deployments regardless of what it would capture.4
UAE summers put ambient temperatures above 45 degrees, coastal humidity is high and airborne dust is constant and abrasive. Consumer-grade hardware does not survive it, which is why sealing is quoted on every specification here.
Sealing across the range runs from IP54 to IP68. Those are not grades of quality; they describe different environments.
| Rating | Protects against | Appropriate for |
|---|---|---|
| IP54 | Limited dust ingress and splashing water | Indoor posts: retail, reception, hospitality, corporate premises |
| IP65 to IP66 | Dust-tight, and low-pressure water jets | Outdoor patrol in ordinary conditions, covered forecourts, event perimeters |
| IP67 | Dust-tight, and temporary immersion to one metre | Sustained outdoor work, coastal humidity, wash-down environments |
| IP68 | Dust-tight, and continued immersion beyond one metre | The harshest posts: marine, industrial, heavy dust, anything that gets hosed |
Scroll the table sideways on a narrow screen.
Drop protection is quoted separately, at two metres and three metres on the models that publish it, with two units carrying military-standard testing at the 810H revision. A drop rating matters more than it sounds: most damage in the field comes from a camera hitting concrete, not from water.
Heat has a second effect that rarely reaches a specification. Image sensors generate more noise as they warm, so a camera running high-resolution encoding in direct sun produces a visibly worse picture than the same camera indoors.8 Units built for this region route heat away from the sensor deliberately.7 It is worth asking how, and worth testing in the month you will actually be using it.
Sealing and optics also degrade faster here than a warranty period assumes. Abrasive dust dulls a lens and works into seals, which is why a maintenance contract covering optics and seals is worth more in this region than an extended hardware warranty.11
Article 44 of the cybercrime law covers recording and transmitting conversations without consent, and legal commentary treats the UAE as an all-party consent jurisdiction for private conversations.1 Separately, SIRA prohibits microphones on the security systems it regulates unless approval has been obtained; that prohibition is written for a fixed installation, and how it reads against a worn camera is a question to put to SIRA for a specific deployment.3
The specification consequence is concrete. Ask whether audio can be disabled in firmware and locked, so it is a policy decision taken once and not a setting each officer can change. Ask whether the device records the fact that audio was on. Both are much easier to arrange before the fleet ships than after.
The camera captures the event. The platform decides whether the footage is worth anything afterwards, and it is the part of the purchase that is hardest to reverse. Four capabilities are worth treating as non-negotiable.
Where footage is hosted matters as much as how it is secured. For anything touching government work, hosting has to meet the Dubai Electronic Security Centre's cloud standard, which is why the major cloud platforms operate UAE regions.6 On-premises remains a valid answer and is sometimes the simpler one.
Footage of an identifiable person is personal data, and once a face can identify someone it is biometric data.2 Anything shared outside the organisation has to have uninvolved faces and plates removed. Done by hand, an hour of footage can take up to ten hours of an analyst's time.9 For a fleet producing hundreds of hours a month that is not a workflow, it is a headcount.
Automated redaction that detects and tracks faces and plates across a whole timeline turns that into minutes.10 If there is any prospect of footage being released to a client, an insurer or a court, ask what the redaction workflow looks like before you ask what the camera weighs.
A specification that assumes a feature is standard when it is chosen at order is a specification that will not be met. Three deserve explicit attention.
| Capability | Availability | What to do about it |
|---|---|---|
| AI face and plate recognition | Listed on one unit and available as an option at order on two others. Not a range-wide feature | Name it explicitly in the tender, and confirm it against the configuration being quoted and not the model |
| Internal storage capacity | From 16 GB to 512 GB, set at order on most models and fixed at manufacture | Size it from retention and resolution before ordering. It cannot be changed later |
| Night vision mode | On at least one unit, infrared or starlight is chosen at order and cannot be swapped afterwards | Decide against the darkest environment the team actually works in |
Scroll the table sideways on a narrow screen.
AI face and plate recognition is the one that causes the most trouble. It is listed on one unit in the range and available as an option at order on two others. It is not a range-wide capability, and a tender that treats it as standard will get quotes that are not comparable with each other.
What is genuinely consistent across the range is narrower and worth stating plainly: every unit is sealed against dust and water, drop-tested, and encrypts footage on the device. Everything beyond that is a specification decision.
Send us the requirement rather than a model name. Tell us the shift pattern, the environment, who has to accept the footage and whether anyone watches in real time, and we will come back with the configuration that meets it and say where two options are genuinely equivalent.
Storage is the cost that surprises people, because it scales with three things at once. Set the four values below for your own deployment and the numbers update as you move them.
3.8 TBincluding a 20% buffer so an open investigation cannot be overwritten
Figures are an engineering estimate for planning, using continuous recording at the selected bitrate plus a twenty per cent buffer so that an ongoing investigation cannot be overwritten. Real consumption varies with scene complexity where variable bitrate is in use. Send us the numbers and we will size the array against the models that fit.
Hardware is rarely the largest number in a body camera programme, and it is the only one that stops after year one. These are the five that decide the total.
We quote against a requirement rather than a model, so the numbers stay comparable when you put them next to somebody else's. Send us the shift pattern, the fleet size and who has to accept the footage.
These questions work on us and on anyone else you are talking to. The useful ones are the questions where a vague answer tells you as much as a precise one.
The questions buyers put to us first. The full set is on our FAQ hub.
1080p is the working floor and it is what most UAE tenders ask for. Across the range we supply, recording runs from 1280 x 720 up to 3840 x 2160.
Higher is not automatically better. Resolution multiplies against retention to size the storage, so a jump from 1080p to 4K across a large fleet is a storage decision as much as a picture-quality one. Choose the lowest resolution that still answers the question you are buying the footage to answer.
Long enough for the shift plus a margin, which for most UAE deployments means a full working day. Published runtimes across the range run from nine to twelve hours, with one helmet-mounted unit reaching thirteen.
The architecture matters more than the headline figure. A fixed cell is simplest and cheapest; a swappable cell extends the shift with a brief gap in the record; hot-swap with a bridge cell keeps recording through the change.
It depends on whether anybody is watching. Cellular earns its cost when a control room can see an incident while it is running and direct the officer. If footage is reviewed after the event, a docked or Wi-Fi camera delivers the same evidence without a SIM and a monthly data plan per unit.
Connectivity cannot be retrofitted, so decide it honestly at specification time. The question to answer is not whether live video would be nice, but who would watch it and what they would do.
Match it to the post. Buying the top of the range by reflex wastes money. Sealing across our range runs from IP54 to IP68.
Indoor posts are well served at IP54. Outdoor patrol wants IP65 or better. Sustained outdoor work, coastal humidity or anything that gets hosed down calls for IP67 or IP68. Drop protection is quoted separately and matters more than it sounds, because most field damage comes from concrete, not water.
It is fleet size, multiplied by hours recorded per shift, multiplied by the bitrate, multiplied by the retention period, plus a buffer so an active investigation cannot be overwritten. The estimator on this page does the arithmetic for your own numbers.
Two things reduce it substantially: H.265 delivers about the same picture in half the data of H.264, and variable bitrate stops a static scene consuming the same as a moving one.
Treat audio as the highest-risk part of the deployment. Article 44 of the cybercrime law covers recording conversations without consent, and legal commentary describes the UAE as an all-party consent jurisdiction for private conversations. SIRA separately prohibits microphones on the security systems it regulates unless approval has been obtained.
In practice audio should be a decision taken deliberately and documented, with notice given at the point of contact. Where a deployment has no clear basis for it, video only is the safer position. Ask for firmware that can lock the setting.
No. It is listed on one unit in our range and available as an option at order on two others. It is not a range-wide capability.
If recognition is a requirement, name it explicitly in the tender and confirm it against the configuration being quoted, not the model. A specification that assumes it is standard will produce quotes that cannot be compared with each other.
That is the last question, not the first, and it is a short conversation once the requirement is settled. Tell us the shift pattern, the environment, whether anyone watches in real time and who has to accept the footage.
We will come back with the configuration that meets it, say plainly where two options are genuinely equivalent, and tell you where a requirement is not published by any manufacturer in the range.
This page is general technical and procurement guidance for body-worn camera deployments in the UAE. Specifications are the published range we supply and are quoted as ceilings and floors; individual configurations vary and several capabilities are set at order. Storage figures are planning estimates. Regulatory summaries are general information and not legal advice. More guides are collected on our resources hub. Confirm any figure with us for a specific configuration, and take advice from a qualified UAE practitioner on the regulatory position.
Send us the shift pattern, the environment, the fleet size and who has to accept the footage. We will come back with the configuration that meets it, and say plainly where a requirement is not published by any manufacturer in the range.
Complete body-worn camera systems for government and enterprise operators across the UAE and GCC.
© 2026 BodyCamera.ae
Two ways to deploy. Browse below.
Tell us where the cameras will be deployed and we will arrange a demonstration with the right models for your sector.
