The State Fiscal Service of Ukraine is actively using artificial intelligence technologies to monitor agricultural land together with the DroneUA company.
In April 2019, the next stage of the pilot project for scanning the Lugansk and Lviv regions using satellite technologies and automatic field classification systems was completed.
Project task – automatic determination of the targeted use of land resources, compliance with crop rotation, determination of the exact boundaries of the cultivated land bank. This is the formation of an analytical and research base in cases of discrepancies between the fact of the fields being processed and the reporting provided.
During this type of research, the method of artificial neural networks (ANN — artificial neural network) is used, a virtual computer environment that, based on a set of training images, detects and outlines the desired objects in satellite images — in this case, fields with a specific crop: wheat, corn, rapeseed, etc. Each crop has its own characteristic dynamics of plant development, so the input to the neural network is a set of Landsat and Sentinel satellite images for the growing season.
To train neural networks, a sample of physically confirmed fields in which drone flights were carried out or visual identification of the culture by DroneUA field team members specifically for this project is used.
As a result, current maps of crops are created with detail from 10 m/pixel with coordinate reference and automatic identification of the crop crop. The accuracy of crop identification exceeds 95%.
Statistical information is provided by district, region and further – at the country level.
During the work in certain areas of the Lugansk region, violations in compliance with crop rotation exceeded 27%, while every third field was cultivated with violations of certain contours and the intended purpose of land resources. The studies were supplemented with physical examinations to verify the accuracy of the results.
Research result – not the formation of an evidence base, but the creation of recommended coordinates for research by employees of the state fiscal service, as well as for tracking the dynamics of violations over time.
«This is our first experience of cooperation with the State Fiscal Service. It's frankly inspiring that government officials are interested in innovative tools like this. We are now entering a new era of big data, where it is impossible to falsify statistical information or information about cultivated areas. We must be prepared for the fact that modern business must be transparent» — says Valery Yakovenko, co-founder of DroneUA
If the assessment is positive, the pilot project will be expanded to other regions of Ukraine in 2019.