Scientists of the Altai State Agrarian University and the All-Russian Research Institute of Phytopathology continue the implementation of the joint project "Development of methods for the timely detection of diseases, pests and weeds in the fields using technical vision and intelligent systems for the transition to the introduction of pesticides in differentiated doses", reports press service of the Altai State Agrarian University.
According to the project plan, scientists will develop methods and technologies for ground and remote detection of pests, diseases and weeds in crops using digital multispectral and hyperspectral cameras and artificial intelligence algorithms.
The team of scientists of the Altai State Agrarian University involved in the implementation of the project is headed by Doctor of Technical Sciences, Professor, Head of the Department of Agricultural Machinery and Technology Vladimir Belyaev.
The key stage in the implementation of the project was the field testing of the design of a vertical optical sensor system with high resolution imaging (on a millimeter scale), with the ability to work at different heights in crops, with parallel recording of the track and coordinates of survey points while moving. The experiment took place on the fields of the industrial partner of AGAU - the farm LLC "Leo" in the Kalmansky district of the Altai Territory, on soybean crops of the Gratsia variety. Scientists from the Research Institute of Phytopathology arrived in Barnaul to participate in the experiment. Sofia Zhelezova and Ph.D., Researcher Evgenia Stepanova.
The system can be mounted on the boom of a trailed sprayer and, when moving at a speed of 15 km/h at different angles to the surface, record video to assess the presence of harmful objects and weeds in crops and accumulate a spectral library of images of harmful objects.
“One of the tasks of the working group of scientists of the Altai State Agrarian University is the development of a universal camera mounting system and its integration with a GPS receiver for working in the field with the ability to record the track and coordinates of shooting points while moving. In particular, we must experimentally determine the optimal camera angle and mounting height, movement speed, the most effective shooting parameters, etc. Now the results need to be processed and analyzed by colleagues from Moscow,” Vladimir Belyaev commented on the preliminary results of the test.
The next step of the project will be the development of algorithms for processing images obtained by cameras in laboratory and field conditions, using neural networks to classify target objects (diseases, pests and weeds) in images.
Based on the results of the survey of crops, maps of the spatial distribution of harmful organisms in crops will be built.
“Based on the results of ground and remote survey of crops and a map of the spatial distribution of harmful objects, it is planned to develop a decision-making algorithm for the use of pesticides in differentiated doses. Next, a prescription file or spraying task card will be created in a format compatible with the on-board computer of the sprayer., - explains Sofya Zhelezova.
Approbation of the method of spraying crops with pesticides in a differentiated dose and a preliminary economic assessment of this method of spraying in comparison with traditional spraying in the same dose over the entire field area is the final task of the project, the scientists add.