Can artificial intelligence help protect horses from mutilation?

The increase in cases of mutilated and killed horses in the French countryside is arousing outrage and anger. These acts, which have not yet been explained, are carried out at night in remote fields that are difficult to monitor for breeders and stud owners. Still. At a time when networked cameras and the Internet of Things (IoT) open up new perspectives far beyond surveillance (analysis of customer trips in shops, counting of visitors or customers, of vehicles for smart cities), compliance with health standards on archaeological Sites, parcel tracking, etc.), the secrecy and impunity with which the perpetrators of these mutilations act can cast doubt while “defense technologies” appear to exist.

Continuation of the article below

On paper, these technologies could be used to secure almost any field. We know how to install thermal cameras, radars and motion detectors, or even combine them with drones. We know how to either wire them or connect them to relays or cellular networks (3G, 4G) to send notifications. We also know how to embed video analysis algorithms in these devices or upload information to a cloud to analyze these flows (in real time or after the fact).

Is this mix of artificial intelligence (AI) and IoT still mature enough to protect horses in France and Navarre? The beginning of an answer can be found on the Canadian site.

Live peacefully with polar bears

On the border of the Arctic, the small town of Churchill borders the Hudson Bay in the far north of the province of Manitoba. The tranquility of this city with almost 1,000 inhabitants is still regularly disturbed by two very different types of visitors.

More than 10,000 tourists come to Churchill each year to see other visitors: the polar bears. Due to global warming, bears are expanding their hunting grounds to find food and are increasingly arriving on the outskirts of the city.

A polar bear is very pretty, but also a deadly danger to humans. The local authorities have therefore developed a sophisticated detection system to warn residents (and tourists) when a bear is nearby.

Until 2016, it was animal welfare officers who called a headquarters when they spotted a bear so that other officials could push it away (without damaging it) or immobilize it and release it in the air. With more than 300 calls in 2016, another route had to be explored in 2017.

“When we analyzed Churchill’s situation, we thought a solution with cameras and radars that we saw at work on the other side of the country a few years ago would be a good idea. »Explains BJ Kirschhoffer, head of field operations at Polar Bears International (an association for the defense of polar bears). The other project he mentions is a deployment that placed smart cameras and sensors around pipelines in Alaska.

The principle at Churchill was to replicate this device to automate the detection of bears thanks to the real-time analysis of the returns of a camera (a Canon VB-R10VE) and two radars through video management software (VMS). Software that uses AI to analyze the data output by these connected devices. If a bear gets too close, the system sends an email or a warning. An agent can then take a hand and view the images from anywhere on a tablet or smartphone.

UI tool for recognizing polar bears in Churchill

(Image credit: Milestone Systems)

“This technology has great potential to help conserve flora and fauna in a number of ways,” said Liz Larsen, nature conservation director at Hogle Zoo in Salt Lake City, a partner of Polar Bears International and a sponsor of the program. “We’re looking at this project [à Churchill] as a promising tool to ensure the long-term survival of animal populations, especially those most vulnerable to climate change. “

For them, this particular type of AI needs to be researched in order to “save wild elephants, hippos and other non-predatory species that frequently invade and destroy crops, threaten humans and provoke defensive reactions in them.” that can be fatal [pour les animaux] “, She explains

But if technology can recognize animals to protect humans, why couldn’t it recognize humans to protect animals?

On paper, but in “real life”?

The answer is that it is possible, but the devil, as always, is in the details (and money).

Return to Churchill’s Snow and Icy Waters. The camera and radars are not installed in the middle of nowhere. On the contrary, they are in one of the busiest buildings in town. This building – near an active polar bear migration corridor – houses the school, hospital and community center. The installation of sensors and their connection is therefore far less problematic than in an isolated area.

Another element, the city is certainly small and a little lost in the vastness of Canada. But it has good cellular coverage and high-speed WiFi. Again, many country estates and farms in France are not in this case. This does not mean that such a device cannot be installed there, but that topographical and network specificities can make this considerably more difficult.

The project that inspired Churchill was in Alaska. It can be assumed (officials do not provide details on the network) that alternatives to WiFi had to be found in this case (e.g. with solutions for wireless data transmission over several hundred meters). So anything is possible. As long as you have the budget. However, horse breeders do not have the same resources as oil tankers from the far north.

“We can well imagine covering areas with radar technologies (LiDAR), defining what an intruder is (so as not to activate as soon as a fox or a dog comes by) and sending warnings to the horse’s owner, who can decide to go to the police or to call ”, confirms MagIT Remy Deutschler, Country Manager France of Milestone Systems, a VMS software manufacturer who works in particular with RATP in France.

“A thermal imaging camera can also be used if you want to cover a relatively larger area. Here, too, the VMS can determine whether a hot spot is a wolf or a person. In some cases there is an interest in assigning a camera for double checking, but it is not mandatory in this context. “

“We can also imagine drones patrolling a certain sector at night,” he continues. “Or cameras that are not necessarily connected, but have integrated software that generates metadata in order to be able to research quickly afterwards.” Specifically, such a tool enables the search for an object (e.g. a red sweater) over a period of time (from 10 a.m. to 10 p.m.). This is where a VMS could find a man at night – avoiding hours of video surveillance, which is useful for gendarmes who don’t necessarily have the time.

“Technology could meet that demand. It exists and is under control, ”assures Remy Deutschler. However, the expert qualifies immediately. “But regardless of the emotional side, which is still very important, there is always the cost-benefit ratio,” he confirms. Because these installations are obviously not free and the benefits are often difficult to assess (what is the price for a status quo?).

Unite or find a sponsor?

In short, IoT, VMS, Cloud, IA, all of these technologies could help protect horses.

You could. In reality, however, they come at a cost: you have to find the right certified supplier, buy the equipment, analyze the site, install it, provide a network, learn the algorithms, train how to use the tool, etc. – which unfortunately they do for ordinary growers makes inaccessible.

They still remain an opportunity to explore – why not? – a pooling of resources (for example within the FFE?) or the search for technological sponsorship with the IT giants (a marketing investment in the form of providing their tools which would be very profitable for them in terms of image). It would be a shame if the equestrian world could only carry out its digital transformation at a very small trot.