How Halter detects heat & more...


Meet Harry our Head of Data Science. We plucked him from NASA and he now oversees some of the most valuable functions you’ve told us you want in the product. Harry uses something called artificial intelligence which you may have heard about.  It is the “brains” in our product, helping us to understand the health and well-being of your cows. Nothing about that sounds straightforward, so we asked Harry to give us a little more to work with...

Artificial intelligence - we hear that term a lot on the news, what is it all about?

Human brains are capable of processing a lot of information and carrying out complex decision making, sometimes without us consciously realising it! Artificial Intelligence simulates this intelligence and decision making in machines . To do this, various algorithms, made by humans, are used to train a computer about how to process complex information for itself.

You mentioned Algorithms? What on earth does that mean..

An algorithm is simply a set of instructions that something follows to complete a task. We as people can also make decisions like algorithms. For example, if the weather is sunny, then you might walk to work, otherwise you might drive in instead. Halter’s central control algorithm, named Cowgorithm™, can monitor a cow’s activity and allows farmers to do smart things such as shift the cows remotely.

So how do you use artificial intelligence to detect heat in cows?

I like to call this product AI for AI (Artificial intelligence for Artificial insemination). Halter can use behavioural data we have collected directly from cows to build machine learning models, and based on information such as mounting behaviour and general restlessness we can generate a heat index that will alert the farmer when cow behaviour differs significantly from their normal behaviour.

Is automatic heat detection the future?

Traditional techniques such as using tailpaint to indicate mounting events for a cow are labour intensive, and also do not catch all events, as 70% of milking cows fall in heat between 6pm and 6am (after stable work hours). In addition, there is also evidence to suggest that cows are having shorter periods of heat with lower intensity, making it harder to detect oestrus manually. Increasing in-calf rate by even a few percent can boost overall milk production drastically.

Wow. So what else can you predict?

With all of the information we can gather from the collar, we can predict many other aspects of cow behaviour such as calving , lameness, sickness, or anything that is out of the ordinary. This is especially useful as we can make actions based on these predictions. If a cow is sick, she can be drafted away from the rest of the herd until she can be attended to.

Now, despite the ‘how’ behind Halter’s heat detection being very techy, rather complicated and somewhat difficult to understand, it’s application is anything but. We’re looking to make farmers lives easier so we transform this data into something simple, an alert. If a cow is in heat, we alert you and if not then we don’t, easy as that! Halter reliably identifies heat so that you get all cows in-calf quicker and spend less time doing it!

Halter’s Head of Data Science - Harry She

Halter’s Head of Data Science - Harry She


We’ve got your questions covered