How Show Jumping Vets Use Thermal Cameras to Catch Horse Abuse

The FEI's veterinary rulebook contains one number that decides nothing at all. A temperature difference of more than two degrees Celsius between a horse's left and right lower limbs counts as a significant thermographic finding, and a significant thermographic finding eliminates nobody, fines nobody, and never becomes the result of record. Everything it can do, it does by handing two veterinarians a reason to keep working the leg with their hands.

That is a rare thing to find written down. Sport governing bodies are not usually careful about the difference between an image that suggests something and an image that proves it, and this one reads like it was drafted by people who know what a thermal camera actually reports. The same reasoning governs the thermal attachment you can clip to a phone for $214, and almost nobody applies it there.

What the Rulebook Actually Says

The limb sensitivity protocol lives in Articles 1046 through 1051 of the FEI Veterinary Regulations, the rulebook the Fédération Equestre Internationale writes for international equestrian sport, in its 15th Edition 2023, effective January 1, 2026. The protocol is not tucked into an annex, and it is not discipline-specific on paper, though the behavior it targets belongs to jumping, where a horse whose lower legs sting is expected to fold them tighter over a fence and stop touching rails. The regulations call the resulting state hypersensitivity. A hypersensitive limb, in the FEI's words, has "an excessive or abnormal reaction to palpation." The mirror-image offense is hyposensitivity, a leg deadened by a nerve block or a neurectomy, which matters most in endurance.

Any horse can be examined at any point during an event, randomly or because someone decided to target it, including between rounds and before a jump-off. Two examining veterinarians appointed by the FEI Veterinary Department run the exam. It consists primarily of palpation, and Article 1047.4 adds that where appropriate, a thermography camera will be used to image all four lower limbs. The rulebook lets that thermography exam come before the palpation or after it. The initial examination is recorded on video, and the thermal images, where they exist, are saved.

Article 1047.6 then defines what counts as a significant thermographic finding, and there are exactly two of them: "A temperature difference between collateral limbs exceeding two degrees Celsius," and a localized or generalized region of high or low temperature in one or more limbs. Separately, Article 1047.7 defines significant clinical findings, which are an abnormal reaction to palpation or other visible changes to the skin. These are two different lists, and the camera cannot generate an entry on the second one.

If both examining veterinarians agree the horse has abnormal limb sensitivity, they tell the rider, who is offered the chance to withdraw the horse without further consequence under that article. Decline that, and the horse goes to a Final Limb Sensitivity Examination, run by a different veterinarian in front of a member of the ground jury. Article 1048.3 describes that final examination as consisting of the palpation of the horse's lower limbs. There is no camera in it. To disqualify the horse, the two initial examiners, the final examiner, and the ground jury member all have to agree unanimously, and the decision is issued within thirty minutes. There is no appeal. Three disqualifications or withdrawals in any twelve-month period suspend the rider for two months, absent exceptional circumstances.

So the strictest consequence in the protocol rests on three veterinarians and one ground jury member agreeing about what was found by hand. The thermal image is allowed to point. It is not allowed to decide.

A Thermal Camera Does Not Measure Temperature

The popular version of thermal imaging is that you point the thing at a surface and it tells you how hot that surface is, to a tenth of a degree, because the number on the screen has a tenth of a degree in it. That version is not stupid. Modern microbolometers really are astonishing at seeing small differences. Seek Thermal quotes 35 mK of thermal sensitivity on its current phone attachments, meaning the sensor can distinguish two points in a scene that differ by about a thirtieth of a degree.

The same spec sheet carries a second number. The Seek Nano 300, at $519, states its measurement accuracy as ±5 °C or 5%. The $529 FLIR Edge Pro does better and still publishes ±3 °C or ±5%, applicable only sixty seconds after startup, with the camera itself between 15 and 35 °C and the scene between 5 and 120 °C. A device that resolves a thirtieth of a degree is allowed to be wrong about the absolute value by five whole degrees.

Both numbers are honest, and together they describe what the hardware is. The sensor measures radiant power arriving at the lens. Converting that power into a temperature requires the camera to assume how efficiently the surface emits infrared, how much infrared from the surroundings is bouncing off it toward you, and how much the intervening air absorbs. Get any of those wrong, and the absolute reading drifts, while the relative structure of the image stays intact.

That is why the FEI wrote a differential rule. Two degrees between one leg and the other, in the same frame, under the same sky, with the same assumed emissivity, is a comparison in which most of a camera's systematic error simply cancels. Near skin temperature, emitted power climbs by roughly 1.3 percent per degree Celsius, so a two-degree asymmetry is a couple of percent of signal, sitting far above what even a cheap sensor can resolve. As a screening trigger, it is well chosen. As proof of anything, it is not offered.

The Things That Move the Number Without Anyone Cheating

Start with emissivity, the assumption the camera cannot verify. Equine researchers generally set 0.98, and a 2024 equine thermography study used exactly that value as its software default. Run the radiometric arithmetic yourself, and the effect is real but modest: a camera set to 0.98, pointed at skin genuinely at 25 °C whose true emissivity is 0.90, against a cold 0 °C background, reports about 23.2 °C. That is a 1.8-degree error, and on a warm day it shrinks to well under half a degree.

If both of the horse's front legs are equally dirty, equally greasy, and equally hairy, that error lands on both of them identically and vanishes from the comparison. Emissivity only manufactures a false asymmetry when the two legs genuinely differ, and legs differ constantly. One gets a liniment or a coat dressing and the other does not. One picks up wet sand from the arena. One gets hosed off and cools by evaporation while the other stays dry.

The larger effects are not optical at all. In a 2014 study in the American Journal of Veterinary Research, ten horses were lunged for twenty minutes with a bandage on one forelimb and a tendon boot on the other. At rest, the two limbs were indistinguishable, 15.3 °C under each. After exercise, sensors fixed to the skin beneath the equipment read 24.8 °C under the bandage against 20.6 °C under the boot. That four-degree spread came from contact sensors under the wrap, and it is not a like-for-like comparison with the FEI's camera threshold, because a thermal camera can only read the limb once the equipment is off. The study measured both ways. What the result establishes is that the asymmetry is real, and that a groom reaching for different equipment on different sides is enough to produce one.

Ambient conditions do the rest. Work published in the Equine Veterinary Journal in 1992 tracked forelimb surface temperature in ponies through controlled changes in ambient temperature and found that limb surface temperature followed it, that it was least stable when the surroundings were cool, and that limb temperature sometimes climbed sharply and spontaneously. The authors concluded that reading a limb's surface temperature clinically requires knowing the ambient conditions it was read in. The equine distal limb is not a passive surface. It is thermoregulatory hardware, dumping or conserving heat on its own schedule, and a horse standing at the gate with the morning sun on its left side is running an uncontrolled experiment on the person holding the camera.

The regulations know this, and they show it by omission. Search the 2026 Veterinary Regulations for an emissivity setting, a required camera specification, a calibration schedule, an acclimatization period, or an ambient temperature window for the thermography exam, and there is nothing. That is not sloppiness. Elsewhere in the same document, pony measuring requires FEI-approved sticks that are calibrated, and noseband tightness requires an FEI-approved measuring device with its own written protocol. When the FEI wants a number that can end somebody's competition, it specifies the instrument. It did not specify one here.

Where the Sport Puts a Real Instrument

Annex VIII of the same rulebook covers hyposensitivity testing in endurance, and the contrast is total. That examination uses a device approved and certified by the FEI, applied up to three times per limb, with a stated numerical threshold: an evasive reaction below 20 Newtons is negative, while no reaction, or a reaction only at 20 Newtons or above, is a failure. A horse that fails is disqualified and sits out an additional 28 days.

One protocol produces a decision from a calibrated force reading. The other produces a decision from four people agreeing about palpation, with a camera used to point them at the right horse. Nobody at the FEI mixed those up. I have ridden all my life, competed, worked in stables, and both announced and judged shows, and the sport's instinct here is better than its public reputation. Everyone at the ring understands that a hot leg is a question rather than an answer, because legs run hot for a dozen honest reasons every day.

The Same Camera Is Now Clipped to Your Phone

This belongs to photographers now because the hardware has arrived in your pocket, and the marketing around it has none of the FEI's discipline. Phone thermal attachments get sold for wildlife spotting, roof and insulation surveys, moisture hunting, and electrical inspection, all of which are jobs where somebody will eventually treat your image as evidence.

A 2024 paper in PLOS ONE tested a phone attachment against a professional instrument on exactly this problem. Researchers imaged horses' limbs before and after anesthetizing the median nerve, using a high-end handheld FLIR P640 and a smartphone-mounted FLIR One side by side. Both devices tracked the same underlying change and correlated well, and the phone device also read warmer overall, produced a wider temperature range, and registered a temperature rise in the untreated leg. The authors' conclusion was that the smartphone device should not be used to assess that nerve block, and that skin temperatures obtained with it should be interpreted with caution.

Getting that far took a laboratory. Every image was made in a climate-controlled room held at 21 °C and 50 percent relative humidity. Horses were groomed sixty minutes before imaging and given twenty minutes to acclimatize. The camera sat two meters from a mark on the floor, every time. Those controls were not ceremonial. They were what it took to make a difference of a degree or two in a horse's leg mean anything, and none of that apparatus exists in an attic, a barn aisle, or a field at dusk.

What survives the move outdoors is comparison. Two identical stud walls in the same frame, one cold and one not. Two bearings on the same machine. Two sides of the same roof, shot in the same minute from the same distance. Shoot differences, keep everything else constant, and say plainly in your caption what you compared. Absolute numbers are the fragile part. A radiometric camera will hand you a defensible one, but you have to earn it: set emissivity for the surface actually in front of you, measure the reflected apparent temperature rather than guessing it, account for distance and humidity where the path is long enough to matter, and stay inside the camera's stated operating window and accuracy spec. Which of those you can safely skip depends on what you are pointed at. A high-emissivity surface a few meters away in still air forgives most of them, and the error you would have corrected is small next to the camera's own accuracy limit anyway. A glossy surface under an open sky forgives nothing, because a clear sky is an enormous cold reflector that a low-emissivity surface will mirror straight into your lens, and a wet surface reads cold for its own reason, evaporating while you shoot it.

So when a thermal frame turns up online as proof of a leak, a fever, or an act of cruelty, run the FEI's test on it. Ask what the camera assumed, ask whether the two things being compared were alike in every way but the one at issue, and ask who put their hands on it afterward. At a horse show, that last answer is already written into the rules. The images get saved, the boots come off, and somebody's thumb runs down the tendon while a ground jury member watches.

Lead image by Clément Bucco-Lechat, CC BY-SA 3.0. Source.