Smart App Could Improve Sunflower Fertility Roguing
Students from Vavilov University and Saratov State Technical University have developed AgroWeed, a startup project designed to automate fertility roguing in sunflower breeding and improve seed quality.
When developing new sunflower varieties, thorough roguing during plant growth is critical. To produce first-generation hybrids, the female parent is deliberately made male-sterile, meaning it produces no pollen, while the male parent is fertile and produces pollen. Hybrid production relies on controlled pollination. If fertile plants appear in the female rows, they will release pollen that reaches the stigmas of male-sterile plants. That disrupts controlled crossing, leads to self-pollination and produces seeds that do not carry the new hybrid combination of traits. The resulting seed lot will therefore be defective, putting the field’s hybrid seed-production system at risk. Breeders consequently carry out fertility roguing throughout the growing process, removing male plants from female rows.
Manual Work Ready for Automation
Roguing is highly labor-intensive because the work is still performed manually. Workers can become fatigued in the field and misclassify plants, increasing the risk of defects in the breeding process. Manual labor also adds time and cost to producing new seed. Automation could address virtually all of these problems, but it requires computer vision systems and a neural network capable of distinguishing female, or male-sterile, plants from male, or fertile, ones. Fertile plants can be reliably distinguished from male-sterile plants by the color of their anthers: Male-sterile plants have pale yellow anthers, while fertile plants have dark ones. These plants must be detected and removed at the beginning of flowering, before they shed pollen.
A team formed specifically for the AgroWeed startup is training the digital platform to perform that task. Students from Saratov State University of Genetics, Biotechnology and Engineering named after N.I. Vavilov (Vavilov University) and Yuri Gagarin State Technical University of Saratov have developed an AI-powered application to automate fertility roguing in sunflower breeding. On August 27, the students worked in a field in the village of Uritskoye in Russia’s Saratov region, collecting data to train the neural network. They took more than 1,000 photographs of fertile and male-sterile plants during the growth stage. Once the images have been annotated, they will be added to the training dataset.
Digital Quality Control for Roguing
AgroWeed is designed to provide quality control for fertility roguing. A mobile app was developed for that purpose. The AI platform determines whether a plant is fertile or male-sterile within two to three seconds and displays its precise outline on the screen. The platform can also operate without an internet connection, analyzing data directly on the device. Once connectivity becomes available, the information is transmitted to a server for more precise analysis. Meanwhile, the model continues to learn from real-world field images.
In the future, the AgroWeed digital platform is expected to become a digital hub for automated equipment used to mechanize sunflower roguing. Integration with autonomous control systems could eventually allow agricultural machinery to perform the work without a human operator while enabling the roguing process to be monitored remotely.
For Russia’s agricultural sector, AgroWeed could mark an important step in modernizing domestic seed production. Automating roguing would accelerate hybrid development and help guarantee seed quality, reducing growers’ dependence on imports of this critical agricultural input. Experts estimate that domestically bred sunflower varieties accounted for 25% of seed planted in Russia in 2022, rising to 59% in 2025. By 2030, the share of domestically bred seed is expected to reach 75%, a target established in the country’s Food Security Doctrine.
At the same time, crop producers could reduce their costs. An analysis of publicly available commercial offers from Russian and foreign seed companies indicates that in 2025, Russian sunflower seed cost 72% less on average than imported seed.
Exporting Russian Seeds and Technology
Digital technologies are increasingly being adopted to advance plant breeding in Russia. Scientists at the All-Russian Research Institute of Agricultural Biotechnology and the Moscow Center for Advanced Technologies, for example, have developed a computer vision system for accurately counting seeds and assessing their quality on harvested sunflower heads. Steppe Agroholding uses machine vision to count sunflowers and assess their quality in the field.
Digitalization could allow Russian agricultural producers to increase production and, in turn, exports of sunflower and processed sunflower products. According to OleoScope, Russian vegetable oil exports increased 6% year over year between January and July 2026, reaching 4.18 million metric tons.
Russian seed exports are growing as well. In 2025, Russia exported seed material to more than 30 countries, with shipments going not only to traditional markets across the former Soviet Union but also to countries in Africa, the Middle East, Asia and Latin America. These regions are key markets for further growth in Russian seed exports. Once adapted to other varieties and climatic conditions, Russian digital technologies for accelerated seed production could also be supplied to countries with developed crop-production sectors that are interested in building their own plant-breeding capabilities.
