Customer stories
Sensand explains how Picterra helps them measure farm-scale environment performance.
Industry:
Agriculture
Business type:
Earth Tech platform for farmers
About Sensand
Sensand has one mission: to put Earth Tech into customers’ hands.
Earth Tech is about using technology to solve big problems, and agriculture is under pressure like never before. With Earth Tech, farmers can increase yields and improve efficiency while reducing environmental impact.
Sensand, an Australian-based business, is building a platform that creates long-term solutions for farmers. They recently engaged with Picterra to assist them with the “Farm-scale Natural Capital Accounting project (NCA)”, which is being run in collaboration with La Trobe University. Sensand’s CEO and Co-founder Peter Moulton explains “We’re working with 50 farmers to zone their farms into management areas for natural capital accounting. The objective is, over time, to look at those zones and see how much they’ve improved environmentally. Ultimately, we aim to show how much natural capital can add to a property’s commercial value”.
Achievements
Challenge
Blockbase, one of Sensand’s key products, is a platform for consolidating agriculture tech and data from different providers which allows farmers to login once and have everything they need to make better decisions. Sensand is currently implementing a classification system for NCA partner farmers within Blockbase. For each of the 50 farms, Sensand uses Picterra to perform a density classification on a current image on part of the property. The analysis provides a starting point for understanding the natural capital state of the entire property. It’s a significant change from how things used to be done. Previously, NCA partners began by physically drawing all farm boundaries manually using GIS software and then needed to classify what sat within those boundaries.
We’re working with 50 farmers to zone their farms into management areas for natural capital accounting. The objective is, over time, to look at those zones and see how much they’ve improved environmentally. Picterra massively reduces the time it takes to do those first pass mapping boundaries. Farmers can simply go in and touch up what the AI model has achieved. That saves farmers a huge amount of time, which has made a material difference to the project’s viability.
Solution
Chandana Kithalagama (CK), who leads engineering on the NCA project at Sensand, explains that Picterra provided a quick way to create functionality for NCA farmers. “Picterra’s fast learning models and intelligent training make our job much easier. These models can be easily customized and trained to a specific use case. For NCA, we can use them, for example, to classify different tree coverage levels on a farmer’s land” says CK.
With the sheer volume of geospatial data now being generated around agricultural properties, there’s already data overload in the farming space — something that Sensand sees as a golden opportunity. “An ability to train custom ML models in Picterra to suit end-user requirements allows us to deliver a solution to a user based on their needs” Peter explains. Picterra allows Sensand customers to create and train their own models, freeing them from relying on product teams to develop solutions.
Benefit
Sensand’s NCA digital product advisor Jim Castles says, “We’re implementing Picterra like a micro-service inside our platform and enabling farmers to use trained models on their data sets. We aim to remove the classification difficulty by using Picterra as an ‘AI model as a service’ that can be trained by non-technical people on the farm. The process is straightforward and to be honest, I didn’t use the training document. I just applied the model over my mum’s farm and taught myself. I’m not an expert, but I improved the model accuracy for the percentage of trees by 5% or so. It’s actually very easy.”
Another way that Picterra is reducing the friction around tech adoption is through licensing: Sensand is on a consumption-based agreement rather than per-user licensing. This means Sensand can effectively let their end -customer just pay a one-off processing fee, rather than needing a commercial licence to work with Picterra. We’re handling the complexity of billing and infrastructure for them. The user chooses their consumption and pays only for that.
Of course, our solution doesn’t remove the need for farmers and consultants to go out on-farm. But it lets them be more precise and focused about where they spend their time when they go out. Using an image to do as much pre-classification work as possible is so powerful. A clear overview of those classifications reduces overhead before farmers start doing the physical verification. The process can be much more targeted, saving consultants’ time, resources, and fuel, thereby reducing costs significantly
Why Picterra
Picterra’s flexibility and ease of use make it much faster and cheaper for Sensand to get solutions to market.
“Because of how our platform is structured, we can very easily bring a lot of different image types into Picterra. That’s everything from WorldView images to users’ own drone imagery. Ultimately, we’re building the system in a way that’s infinitely flexible to whatever the partner’s use case is. So, if a different partner want to use an AI model to count livestock (such as Cownter), we can just swap out the density classification model over to something that suits their use case. This way, we can solve a different problem using the same infrastructure” explains Peter.
