Geospatial machine learning in agriculture & regenerative farming

Shattercane (Sorghum bicolor) and Johnsongrass are common weeds in the United States, harmful and difficult to control. They grow quickly, resemble a corn plant, and compete aggressively with the crop. What makes things worse is that they can introduce fungi and bacteria and are easy to transfer to other fields. Eagle Eye Imaging, an American aerial imaging company, has used Picterra to detect Shattercane and Johnsongrass from drone imagery. What used to require days or even weeks of manual processing, with Picterra, you can do in less than an hour and with high accuracy.

Many growers have already embraced unmanned aerial vehicles (UAVs) like drones as a way to monitor their plantations. This type of Earth Observation (EO) imagery is used to assess plant health, weed detection, weed pressure, pest infestation, drainage and irrigation issues, plant count in a field to access harvest income, etc. Download our ebook to learn how geospatial machine learning helps farmers count objects and segment areas of interest across vast land holdings.

Grapevines represent today a total surface of more than 7 million hectares across the globe. Unfortunately, they’re particularly sensitive to weather and prone to diseases. Vineyards need to be continuously monitored, which is not an easy task as vines evolve quickly throughout the growing season. It’s also difficult to compare them over large domains. Winegrowers need to quickly spot anomalies, disease outbreaks, and plants with not enough nutrients. Luckily, you can do it all with Picterra, without writing a single line of code.

Having up-to-date information about a population of cows or other farm animals is essential for livestock farming. At the same time, counting them has always been a tedious, error-prone and repetitive task. Fortunately, farmers can take advantage of aerial imagery that lets them show the exact location of the cattle. Picterra makes even a bigger difference as it allows for automating detection and counting by farmers themselves – building and training detectors is fast and easy, there is no need to hire programmers or data scientists. An interesting example is a project Cownter, developed by FarmDoctors, Queensland Government, ManGoesMapping and Picterra. They have built a detector specifically for Australian landscapes that often consist of a wide range of plants and soil types.