STRIXE
Menu

Now on Google Play

Turn a phone into something that watches.

A spare Android phone already has an eye and an ear, doing nothing in a drawer. Strixeo makes it an instrument that observes a place — and a thermal camera gives it a third sense it was never born with.

It photographs what visits, names the species on the phone itself, and tells you while the animal is often still standing there. Android 8 or newer.

Get it on Google Play Ask about testing How it perceives What you need

Still being tested at real feeders, and said here rather than discovered later. Writing gets you a reply from a person.

A phone in a black weatherproof housing strapped to a cut post, with a small thermal camera on a bracket beside it, aimed into a clearing. Autumn birch and spruce behind; rain beaded on the housing.
The feeder, 22 October 2026. The phone watches from its housing; the thermal camera is the small box on the arm.

One sighting, end to end

A fox, at eight in the evening, named before it left

This is the whole of what Strixeo does, once. Everything below is how it does it, and where it stops.

  1. 20:07, 27 September. Something warmer than the wood crossed the frame. No light, no flash, nothing for the animal to notice.
  2. It was followed, not merely spotted. One warm shape tracked from frame to frame, so several photographs of it are one visit rather than several arrivals.
  3. Named on the phone itself. The thermal frame was cropped to the body and read by a camera-trap model running on the handset: red fox, 93 %. Offline, in about three seconds.
  4. The pocket knew. One notification, carrying the picture and the name, while the fox was still in the clearing.

A whole night, unedited — including the two it could not name →

A thermal frame of a fox walking past a tree trunk, bright white against black ground, with ears, legs, brush and the line of its back all distinct.
The frame the name came from. Nothing was enhanced — a fox is simply warmer than a wood at eight in the evening.

On your phone

Your side of it

The feeder does the watching on its own. This is all you hold: a list that fills overnight, and any one of them opened full size.

And when waiting will not do, Look now wakes the feeder and puts the thermal picture on your phone live, five frames a second, while you stand there. The camera closes again the moment you stop watching — which is the whole reason a feeder lasts days rather than hours. How the live view works →

The sightings list, showing both named and unnamed visits with their times.
Everything, in order
Named and unnamed together. “Something at the feeder” is what an honest report looks like when the animal was too small to name.
The same list filtered to named sightings: two red foxes, a human and domestic cattle.
Named only
One tap narrows it to what was identified — red fox at 93 %, cattle at 94 % — with the confidence beside every name.
One sighting opened full screen: a person walking away through a wood, bright against cold trees.
One sighting, opened
The frame as the camera wrote it, under the name and the minute it was taken.

What you need

An Android phone you already own, and power. That is the whole of it to begin with. A thermal camera is the one thing worth adding, and it is the sense a phone does not have.

What you need, and where to put it →

Three senses

It perceives in three ways, and they do different jobs

None of them is the whole answer. Sound notices first and never names. Heat sees in complete darkness and names an animal that comes close enough. Sight names the most, and needs light to do it. What each one is for — and where each one stops — is below.

By sound

It hears what no camera can see

A microphone and a 521-class sound model, listening in under a millisecond per window. Hooves on frozen ground, a scuffle in leaf litter, a bark beyond the treeline — sound carries far past anything a lens can resolve, so it is usually what notices first.

Where it stops. Sound wakes the camera; it does not name the animal. Asked directly, the model calls a fox a dog and a roe deer a turkey. The photograph does the naming, always.

Measured: under 1 ms per window, running continuously on the phone

What we support, and what we have tested →

?

By sight

It is the only sense that can say which animal

The phone’s own camera, at full resolution. A small detector puts a box around the animal and a camera-trap classifier names what is inside it — on the device, offline, with the photograph never sent anywhere to be recognised.

Where it stops. It needs light. A visible flash reaches three to ten metres and some animals dislike it. And an animal must be reasonably large in frame: below about a twentieth of the picture, nothing is named at all.

Measured: 30 of 31 named correctly — one test set of 31 infrared trail photographs, not a general accuracy figure

What we support, and what we have tested →

By heat

It sees a body in complete darkness

A thermal camera reads the temperature of everything in front of it, so a warm animal stands out against cold ground with no light at all — through full darkness, fog and light rain, with nothing for an animal to notice.

Where it stops. Size, not darkness. The thermal frame is offered to the species model like any other photograph, and it answers when the body is large enough: nothing under a twentieth of the frame, about three quarters above two fifths of it. Smaller than that and you get a warm shape with its temperature, unnamed.

Measured: a deer at roughly 30 m, 16.6 °C above its background, no moon

What we support, and what we have tested →

What it does with them

Four steps, the same every time, whichever sense noticed.

1

Notice

A sound, a movement, or a warm shape. The camera stays shut until one of them says otherwise. Measured on an Oukitel WP61 (76,800 mWh) at a feeder, screen off: 193 mW listening, about 16 days; with a thermal camera streaming, about 4 days. Cold, signal and a busy night all cost more.

2

Capture

One photograph at full resolution, and on a thermal camera the temperature of every pixel alongside it.

3

Recognise

Find the animal in the frame, then name it — on the phone, offline. If it does not know, it says so rather than guessing.

4

Tell you

One notification for the visit, with the picture and the name, while the animal is very often still standing there.

Honestly

What it will not claim

This is a market that runs on exaggeration, and exaggeration is how you end up distrusting a thing that was mostly right.

So the limits live here, gathered in one place rather than buried one per page. If we tell you plainly where it stops, you can believe us about where it works — and that is the only kind of confidence worth having in something you leave alone in a wood all night.

  • Sound never names an animal. It says something was there and what it sounded like, and a sound label is never presented as a species.
  • Heat names only what comes close. A thermal frame goes to the same species model the ordinary photograph does, and gets the same answer when the body is large enough in frame — the red fox and the cattle on How far it sees were both named from heat alone. Under about a twentieth of the frame it names nothing, and we report a warm shape and its temperature rather than a guess.
  • Range is optics, not ambition. How far it works is set by the lens, not by the software, and detail falls away with distance faster than people expect. Near the limit an animal is a few warm pixels — enough to know something is there, never enough to say what. We would rather tell you which of those two you are getting than quote a headline distance.
  • Warm ground can still look like an animal. Rock and bare earth hold the day’s heat for hours. A warm shape now has to move before it is photographed, which removed most of this — but the threshold is provisional and an evening in open country can still catch us out.
  • Thermal degrees are relative, not absolute. The camera has no shutter and its zero drifts over a night. The difference between an animal and its background is reliable; the bare number is not.
  • It misses things. An animal too small in frame is reported as something warm, unnamed. We would rather tell you that than invent a species.

Why we are doing this

Most of us could not name what lives behind the house

A century ago that knowledge was ordinary. Today it has quietly gone — and the animals have not. Foxes raise cubs behind supermarkets, badgers dig under garden fences, boar walk suburban streets at four in the morning. Wildlife adapted to us far better than we noticed. We stopped looking, and forgot how.

That matters beyond sentiment. People protect what they know about, and nobody argues for an animal they have never seen.

The tools on offer are worse than the problem deserves. A trail camera is a box that stores pictures you look at in three weeks. It does not know what it saw. It cannot tell you while it still matters. And past the few metres its flash reaches, the night it was bought for is invisible to it.

Equipment that does better exists — thermal imaging, species recognition, alerts that arrive while the animal is still standing there. It costs thousands, and it is built for people whose job this is.

Meanwhile there is a small supercomputer in your drawer. The phone you replaced two years ago has a camera, a microphone, a processor that runs neural networks, and memory enough to hold a species model. Somebody paid six hundred euro for it and it is worth forty today. An industry with no interest in wildlife solved the hard part years ago, and nobody pointed the result at a wood.

So Strixeo is not a device we sell you. It is software that takes the computer you already own, gives it a sense it was never born with for the price of a decent lens, and asks it to watch a place and say honestly what came.

Honestly is the part we will not trade. The best wildlife camera in the world is worth nothing if nobody owns one — and one that tells you what you hoped to hear is worth less than none. So we publish what it cannot do beside what it can, and every number on this site is one we measured ourselves.

Experiments — not part of the product

What we are trying next

Nothing here is something you can use, buy or rely on. It is what we are trying, published because it is more interesting than silence. No dates, because we have never given one.

An alert where there is no signal

A small radio at the feeder and another at the house, on the 868 MHz band. It would carry a cue and never a picture — a few bytes saying something is at the feeder — because that is all that fits. It needs hardware at both ends, and it is pointless anywhere there is already a signal, since an ordinary alert is better in every way.

How far it reaches in trees, we have not measured. Published figures for this kind of radio range from a few hundred metres to several kilometres depending almost entirely on height and foliage, so there is no number here until there is one of ours.

A thermometer that means something

How cold was it last night? Nothing in a feeder can answer that today. Phones stopped carrying air thermometers about a decade ago, and the battery’s own sensor reads several degrees above the room it is sitting in.

The thermal camera cannot stand in for one either: it has no shutter, and on one untouched scene its zero moved 3.8 °C in nine minutes. The difference between an animal and its background survives that; the bare number does not. A four-euro sensor beside the radio would answer it properly.

Thermal from the air

The same camera under a small drone, to find out whether a field can be swept rather than watched. There is a little more about it on What you need — including that we still owe it a photograph.

Watch a night, or ask about testing

Strixeo is being tested at real feeders in Orava, Slovakia. The nights go up on YouTube — the ones where it named a fox, and the ones where it named nothing.

Get it on Google Play Watch on YouTube Ask about testing

Free, no account, nothing to buy. Write and a person will read it.