Clever weighing lands on farms

Accurate cattle weights have long mattered to beef farmers, but getting them has often meant extra time, labour and stress on stock.

From left, Scanabull founders Paul Sealock, Dan Bull, Daniel Stuart-Jones and Ursula Haywood.

Now a new weighing technology being trialled around the country is moving from futuristic concept to something farmers, vets and processors can look at in practical terms.

Scanabull has developed a system that estimates liveweight in about a second from a 3D scan captured by either an iPhone Pro or a fixed camera unit called the Scanabull Weigh Point.

The company says the scan is processed on either device using proprietary neural networks so farmers do not need coverage or cloud connections in the paddock.

The approach aims to replace much of the guesswork in buying selling and drafting decisions by giving quick objective weights without yarding stock and running them through a crush.

The startup was founded in April 2024 by Dan Bull who grew up on a farm and later managed livestock.

He has since been joined by co founders with backgrounds in farming veterinary science and artificial intelligence.

Trials and demonstrations are underway with industry partners across NZ including Silver Fern Farms as Scanabull also explores opportunities in Australia and other major beef regions by year end.

Bull says many cattle are still bought and sold on visual estimates and when those estimates miss the mark the financial impact can hit farmers traders and processors.

He says the goal is to give the beef sector a fast reliable way to understand what is happening with animals in real time so decisions can be made on measured performance rather than assumption.

The phone app offers quick spot checks in yards pens or near gateways while the Weigh Point is intended to capture regular pass by measurements in the paddock for trend tracking.

The system generates a detailed point cloud of the animal then a deep learning model analyses body shape and structure to estimate liveweight with results displayed on the device within roughly a second.

Scanabull reports that accuracy on individual animals is already over 93 per cent and that mob level accuracy is higher again with performance improving as more scans are added to the training set.

The technology has been trained using over 100,000 animal data points, enabling it to continuously improve as more livestock scans are captured.

Running this level of machine learning directly on a mobile device required significant technical innovation, the founders say.

Because farms often operate in remote areas without reliable connectivity, the system was designed to run entirely on device using edge computing rather than relying on cloud infrastructure.

“No one is running neural networks at this speed on edge devices for livestock applications. Scanabull has developed our own proprietary architecture and training models to make it possible,” explains Bull.

Co founder and chief commercial officer Dr Ursula Haywood who worked for a decade in rural veterinary practice before moving into agritech says fast objective measurements can help farmers manage animals more effectively and choose better timing for sales.

Haywood says the same capability can provide useful transparency for livestock traders veterinarians and processors who often have limited reliable data on animals before they reach the plant.

Scanabull announced a NZ$1.1 million raise led by Sprout Agritech with support from Enterprise Angels and Callaghan Innovation’s Deep Tech Incubator programme.

Investment will fund further development of Scanabull’s models additional data collection and preparation for broader commercial rollout after the current phase of trials and demonstrations.

Future versions are planned to extend beyond liveweight to include measures such as carcass weight prediction and body condition scoring so more of the supply chain can act on consistent on farm data.