Indoor TV with a camera: how the screen changes what it shows in each aisle

The concept: a screen that answers the aisle

A supermarket has dozens of sections and an absurd number of items competing for the same glance. Ordinary digital signage solves that by anticipation: someone builds the schedule and decides that, from 10am to noon, the cleaning aisle shows the fabric softener promo. It works, and it is the bread and butter of retail signage.

The next layer inverts the logic. Instead of the schedule deciding, shopper behaviour decides. A motion-detection camera watches the aisle, which has been divided into areas during setup — this stretch is paper towels, this one is toilet paper, this one is napkins. When the system detects people lingering in one of those areas, the screen at that spot swaps the slide and starts showing what matters to that section: the item close to its expiry date, the promo of the day, the highest-margin product.

The practical difference is timing. A scheduled grid talks to people walking past; a reactive screen talks to the person who stopped. And whoever stopped in front of the shelf is exactly the person still making up their mind.

The example from the podcast

The scenario came up in the "Podcast Tv Indoor 1" episode, from Café & Tech, when the conversation reached retail. The example given live was as simple as it gets: someone stops in front of the mugs section and, at that moment, the screen at that spot starts showing the mug promo. No operator, nobody watching the floor through the camera to change the content by hand.

The context of that passage was a very real supermarket problem: stock approaching its expiry date that has to move. Instead of the manager relying on a shelf poster and good luck, the screen pushes that item precisely to the person standing right in front of it.

It is worth being honest about the nature of the source: this is a scenario described in a conversation, not a product specification. The podcast named no vendor, no camera model, no detection technology and no result figures — and this article will not invent any of them.

What is real today and what is a custom project

Here is the part most "smart screen" content leaves out, and it deserves to be spelled out.

What JMV Indoor delivers ready and immediately is the layer underneath, which is solid and already covers most cases: cloud indoor TV on an ordinary smart TV, playlists and scheduling per screen, time-based grids, remote content updates. The camera layer is an integration on top of that, and integrations get scoped and quoted — they are not bought in two clicks.

If your operation genuinely has the problem the camera solves, the route is a conversation about feasibility — store layout, number of screens, network, whatever cameras are already on the ceiling — not an assumption that subscribing to a plan is enough.

A camera is not the same as a fixed zone

This confusion is common and expensive, because it leads retailers to price a computer vision project when what they need costs a fraction of that. We covered the zone approach in indoor TV by store zone; the table below separates the two.

They do not compete either: the camera is a layer on top of a zone operation that already runs well. A retailer who could not keep three playlists per screen up to date will not extract value from a system that swaps content ten times an hour — they will end up with ten stale slides instead of three.

One detail that changes the maths: reactivity is only worth anything if there is something to offer at the right moment. It is the same logic as upselling triggers on indoor TV, with one important difference — there the trigger is context (the hour, the queue, the spot in the store); here the trigger is detected physical presence. The offer itself, the discount, the bundle, the product that has to move, is still human work. No camera invents a promotion.

The cheap shortcut before the camera

In that same part of the conversation, before the camera came up, a far simpler solution appeared for the problem that actually hurts day to day: how slow it is to refresh the content of one specific aisle.

The route is the phone shortcuts app — Shortcuts on iPhone, and its Android equivalent — configured to send files straight to the platform's FTP. In practice:

1. A shop-floor employee spots the product that needs to move.
2. They record a short video right there, on their phone, in front of the shelf.
3. They tap a button and the shortcut sends the file to the FTP.
4. The video shows up in the dashboard library and goes into the schedule for that screen.

It is not footfall-reactive and it does not replace the camera — but it solves the speed bottleneck at essentially zero cost and with no project at all. It pairs well with managing your indoor TV from your phone: whoever approves content does it from wherever they are, with no trip back to the office. Honest requirement: someone has to enable and configure FTP access, and the uploaded video still goes through the usual curation before airing — this is not unreviewed auto-publishing.

The market category behind it

This is not one vendor's invention. It is a recognised category within the digital signage market, usually described as targeted digital signage or interactive digital signage, and built on computer vision techniques. The range of implementations goes from very simple to very invasive:

• Presence sensor: the screen wakes from standby when someone approaches. Cheap, nearly trivial, and barely a camera at all.
• Footfall counting and heat maps: how many people walked past and where they stopped, with no individual identification. This is the band the podcast scenario falls into.
• Profile estimation: approximate age bracket and gender used to pick the slide. This already touches sensitive data and needs serious legal care.
• Facial recognition: identifying individuals. A different tier of risk, cost and legal obligation — and rarely necessary to sell mugs.

The first three steps cover nearly everything retail actually wants. The fourth is where projects tend to get tangled up for no good reason.

The part nobody mentions: data protection

A camera pointed at shoppers is a legal matter, and it is worth settling before the technical one. In Brazil, Law 13.709/2018 — the LGPD — governs the processing of personal data; in Europe the equivalent is the General Data Protection Regulation, and most countries have their own version. The common thread: the image of an identifiable person is personal data, and biometric data usually sits in a special category with much stricter requirements.

Where your project lands on that spectrum depends entirely on what the camera does:

• Anonymous footfall counting, storing no images and identifying nobody, is the lightest scenario — though it still calls for transparency with shoppers.
• Age or gender estimation changes the picture, because it starts inferring characteristics about a person.
• Facial recognition is the far end, demanding a robust legal basis and an impact assessment.

Three questions that need written answers before anything is switched on: what is the declared purpose of the processing, what legal basis supports it, and what is retained — for how long and who can access it. Add visible signage telling people the space is monitored. And if the plan is to reuse the security cameras already on the ceiling, take note: they were installed for loss prevention, and using the same footage for shopper behaviour analysis is a different purpose that needs its own justification.

Where to actually start

If after all of that the scenario still makes sense for your operation — long aisles, many sections, purchase decisions made at the shelf — the realistic route is:

1. Make fixed zones work first. Playlists per screen, time-based grids, content kept current. If that is not running smoothly, a camera only multiplies the problem.
2. Measure what you already have. Turnover by section, basket size, items that stall. Without that baseline there is no way to tell whether the new layer paid for itself.
3. Survey the infrastructure. What cameras exist, at what angle and resolution, whether the recorder allows access from another system, and what shape the store network is in.
4. Settle the legal side before the technical one. Purpose, legal basis, retention, signage. It is cheaper to find a blocker in a meeting than after installation.
5. Only then scope the project, with a written scope, one pilot store and a success criterion agreed before you start.

The essentials, in short: a screen that changes by itself according to the aisle is a real idea, with a name in the market and existing cases — and it is, at the same time, a bespoke project with a legal layer, not a switch to be flipped. The good news is that almost all the value it promises starts showing up a few steps earlier, with what is already available today. Questions about how we handle data on our side are answered in our Privacy Policy.

Talk to the JMV Indoor team about your project →