Artificial intelligence is taught to identify areas of effective agricultural activity

Brazilian scientists have trained artificial intelligence to analyze satellite images that show areas where crop production and cattle breeding are effectively combined. The latter concept is considered revolutionary and can potentially ensure the stable development of the agro-industry. But scientists want to know how effective this idea really is.

Artificial intelligence analyzes the state of agriculture from satellite images. Source: Inácio Thomaz Bueno

Integrated system of agricultural activities

Brazilian scientists have published an interesting work in the field of satellite agricultural monitoring. Spacecraft in orbit have been used for a long time to assess the condition of agricultural land. However, in the new study, they have been used to monitor those territories where the system of integration of crop production and animal husbandry is being implemented. 

The crop and livestock integration system, which is called (CLI), is a promising new technology that can change agriculture, although in fact the ideas underlying it are extremely old. They consist in a complex alternation of the cultivation of agricultural crops with the breeding of livestock, mainly cows and pigs.

The idea is that growing soybeans, corn and other similar crops at high yields is much more profitable than livestockHowever, this very crop is actually extremely unstable, including due to the fault of the plants, which severely deplete the land. 

On the other hand, livestock still requires all of the above. In addition, replacing sowing for one season with grazing contributes to land reclamation. Sophisticated modern mathematical models have been invented for all this, and it has shown itself well in experimental areas.

Artificial intelligence and computer images

The only problem is that we still need to understand how the new technology shows itself in mass use and in different conditions. There are more and more agricultural producers who try it every year, and it is almost impossible to observe their results on Earth.

And this is where satellites come into play. In order to monitor the condition of the fields, they have been used for more than one year. The only problem is to find, among numerous images, exactly those areas that are used under the program for the integration of crop and livestock production.

It would be impossible to do this manually, but there is already a proven solution here: neural networks. Therefore, Brazilian scientists could only teach artificial intelligence on ready-made decoded images. In the end, it learned not only to distinguish the necessary areas, but also to assess their condition.

The research site was located in the states of Sao Paulo and Mato Grosso. Object-based image analysis was carried out at intervals of 10 and 15 days in four stages: obtaining CLI data through Planetscope, a group of satellites that capture high-resolution images of the Earth’s surface, showing changes in areas over time; training of algorithms for recognizing CLI-related patterns; displaying CLI areas; and assessment of the accuracy of the model by comparing the automatic results with previous data.

According to

Follow us on Twitter to get the most interesting space news in time