↓ G11 Media Network: | ChannelCity | ImpresaCity | SecurityOpenLab | GreenCity | Italian Channel Awards | Italian Project Awards

Open source tools for Smart Agriculture get even smarter

With Project FarmVibes, Microsoft is releasing in open source new tools and modules data scientists and developers can use to create new applications for precision agriculture


Microsoft has released in open source FarmVibes.AI, a collection of artificial intelligence models that can be used in Smart Agriculture, for the development of advanced data analysis applications. FarmVibes.AI is part of larger project - Project FarmVibes - the components of which will all be open sourced and made available to researchers, data scientists and developers.

The goal, Microsoft explains, is stimulating the growth of data-driven agriculture. An approach where artificial intelligence turns data collected - literally - in the field into useful information to improve farmers' operations. For example, what fertilizers to use and where depending on the type of crops, their condition and weather conditions. Or, where and at what depth to plant new seeds, depending on soil composition.

Data-driven agriculture promises to increase crop production in a way that is adequate to the growing needs of the population but at the same time sustainable. Any improvement introduced by artificial intelligence, also, has a double significance. Agriculture is among climate emergency main causes, so making it more sustainable is positive not only for the market but also for the planet.

FarmVibes.AI currently includes four specific AI algorithms. Async Fusion analyzes data collected from sensors in the soil and combines them with multispectral aerial surveys made with drones or satellite imaging. Analyzing, combining and comparing all these data, the algorithm can display detailed maps of important parameters' distribution, such as nutrient concentration or soil moisture. DeepMC is an algorithm for precision weather forecasting. It analyzes data from sensors in fields and combines them with traditional public weather data, to compute forecasts that are much more targeted and useful for farmers.

SpaceEye is a "supporting" algorithm that applies machine learning to "erase" clouds in satellite images, thus giving a more useful view of the environment the farmer wants to monitor but cannot reach with low-altitude drones. FarmVibes.AI includes also a simulation tool that estimates if and how much specific growing techniques make the amount of carbon dioxide sequestered in the soil vary. Ideally, farmers should use techniques that do not make CO2 enter the atmosphere.

A wider project

Project FarmVibes has three other components that will be released in open source. The most important one is probably FarmVibes.Edge: a computing device that, with the form factor of a conventional PC, has everything needed to perform the conventional processing and artificial intelligence and computer vision algorithms for Smart Agriculture. It is basically an edge computing node designed for farms, that cannot always be connected to the cloud.

Connectivity in large farms is an issue Microsoft addressed also with FarmVibes.Connect. It enables radio coverage using free TV spectrum frequencies, that a farmer can use in many countries with no need of special permissions. FarmVibes.Connect enables both "WiFi-style" broadband coverage for general connectivity and narrowband for long-range IoT applications.

The last - for now - component of Project FarmVibes is FarmVibes.Bot, a platform to develop AI-based chatbots for farmers. The platform is aimed primarily at organizations that need to "talk" with small farmers, geographically dispersed and with almost no IT equipment, or none at all. FarmVibes.Bot is already used, Microsoft explains, by more than 500,000 farmers in sub-Saharan Africa.

Why this matters

Smart Agriculture is a growing and very interesting market. And it's something we can't do without if we want to solve a global food crisis that is becoming worse every year. But innovative startups and small-medium companies aiming at this market face a barrier to entry: developing the necessary AI and machine learning tools and software components is a long and costly process. Too long and too costly, for many companies who want to stay independent and not being acquired after a few demo projects.

Open source can help. Especially this kind of open source, because Microsoft's tools have been tested in the field for a long time. And are now used in real projects by real farmers. While adopting open source components is just a first step in developing a complete Smart Agriculture solution, is a welcomed first step nonetheless. Software developers and data scientists can now build on a solid infrastructure base, focusing their attention on other value-added components.

If you liked this article and want to stay up to date with news from InnovationOpenLab.com subscribe to ours Free newsletter.

Related news

Newsletter signup

Join our mailing list to get weekly updates delivered to your inbox.

Sign me up!