Russian scientists find a way to minimize the use of pesticides
Russian scientists find a way to minimize the use of pesticides
Researchers at the University of Tyumen have created software that can automatically detect and measure the amount of pesticides deposited on plants. The project will help to minimise the quantity of the substances used, lower expenditure and reduce environmental pollution. The findings are published in the journal “Agriculture”.
Pesticides are chemical or biological substances (or mixtures of substances) designed to control harmful insects, rodents, weeds, plant and animal pathogens. Pesticides may also be used for growth regulation, leaf removal or desiccation. Improving the quality and quantity of the crop requires these features.
Pesticides are essential for plant protection. Their proper application is a current research challenge. According to scientists, organosilicon adjuvants are becoming more prevalent in modern agriculture. They are added to the pesticide solution, enhancing its effect on the pest. However, overuse of these substances can be dangerous for insect pollinators and affects the biosafety of the cultivation region.
The researchers utilised a neural network to measure the area of plants treated with pesticides. The neural network can identify dry zones of the leaf that are not covered with the substance. Usually these dry areas are counted in the total coverage area, and this gives incorrect information on efficiency.
The scientists' project is ready for application. To put it into practice, all that is required is a laptop computer and a camera. Neural networks only need a few seconds to study each image.
Scientists around the world are exploring the future applications of artificial intelligence. However, there are not many examples of AI applications in image processing in agriculture. Researchers at the University of Tyumen have adapted an existing artificial intelligence model used for automatic recognition of satellite images.
Fabio Grazioso works at the Photonics and Microfluidics Research Laboratory of the University of Tyumen. He noted that the quality of artificial intelligence inference is highly dependent on the process of training the neural network. Members of the research team manually recognised 130 images of leaves and then used them to train the AI. Fabio Grazioso added that the results could be significantly improved by conducting a longer training process with more images.
The research is a part of the state task of the Ministry of science and higher education of the Russian Federation.
Pesticides are essential for plant protection. Their proper application is a current research challenge. According to scientists, organosilicon adjuvants are becoming more prevalent in modern agriculture. They are added to the pesticide solution, enhancing its effect on the pest. However, overuse of these substances can be dangerous for insect pollinators and affects the biosafety of the cultivation region.
The researchers utilised a neural network to measure the area of plants treated with pesticides. The neural network can identify dry zones of the leaf that are not covered with the substance. Usually these dry areas are counted in the total coverage area, and this gives incorrect information on efficiency.
The scientists' project is ready for application. To put it into practice, all that is required is a laptop computer and a camera. Neural networks only need a few seconds to study each image.
Scientists around the world are exploring the future applications of artificial intelligence. However, there are not many examples of AI applications in image processing in agriculture. Researchers at the University of Tyumen have adapted an existing artificial intelligence model used for automatic recognition of satellite images.
Fabio Grazioso works at the Photonics and Microfluidics Research Laboratory of the University of Tyumen. He noted that the quality of artificial intelligence inference is highly dependent on the process of training the neural network. Members of the research team manually recognised 130 images of leaves and then used them to train the AI. Fabio Grazioso added that the results could be significantly improved by conducting a longer training process with more images.
The research is a part of the state task of the Ministry of science and higher education of the Russian Federation.