Digital Phytopathologist Set to Protect Crops
Researchers in Perm have developed an AI-based system for detecting almond diseases, achieving accuracy levels of up to 90%.
Today, one of the most in-demand professions in agriculture is that of a phytopathologist – effectively a doctor for plants. These professionals study plant diseases and determine how to treat them. In practice, they decide how to handle infected fields, which treatments to apply and what crops should be planted after an outbreak. During disease outbreaks, their role also includes protecting neighboring fields from infection.
Each year, weeds, pests and plant diseases destroy an estimated 25% to 40% of potential global crop yields. As a result, managing plant health remains one of the most critical challenges in agriculture. Even a modest reduction in losses could unlock significant additional food resources.
Protecting Yields
Due to a shortage of phytopathologists, many farms still rely on visual diagnosis, with agronomists identifying diseases based on external symptoms. This process is time-consuming, and diagnostic errors often lead to crop losses and unnecessary spending on ineffective treatments. Automated diagnostic systems offer a more efficient way to address this challenge.
Researchers from Perm National Research Polytechnic University and Perm State Agro-Technological University have developed a digital solution that uses AI to analyze plant images, identify diseases and recommend treatment strategies.
Like any neural network, the system needs training to perform effectively. Developers began with almonds, a crop with growing importance for Russian agriculture, as about 90% of nuts on the domestic market are imported. Efforts are underway to expand local production, with almond orchards increasing annually by 200–300 hectares in Crimea and new plantings emerging in Krasnodar Krai and Kabardino-Balkaria. However, protecting the crop from disease remains essential.
A Capable AI Doctor
The new neural network has demonstrated strong performance as a diagnostic tool for almonds. It is designed to be simple and practical in real-world use.
“To use the system, a farmer simply takes a photo of a suspicious leaf using a mobile application. The neural network analyzes the image, evaluating leaf texture, color, shape and the presence of spots. Based on these parameters, it determines the disease. If a diagnosis is confirmed, the system provides treatment recommendations,” said Sergey Kostarev, Doctor of Technical Sciences and Professor at the Department of Information Technologies and Automated Systems at PNRPU.
Field validation in Crimea showed that the AI system diagnoses diseases with an accuracy of 70% to 90%. More importantly, the use of the system reduced seedling mortality to just 1%–2%, significantly improving outcomes for growers.
The platform will be adapted for other crops grown in different regions of Russia. Expanding the dataset will enable continuous learning, making the system increasingly versatile and applicable across a wide range of agricultural operations.
In parallel, similar AI-driven plant diagnostics solutions are being actively developed across Russia. Researchers at Tyumen State University have created an intelligent system for monitoring plant health in greenhouses. Developers in Dubna have launched an online platform for plant disease detection and the DoctorP mobile application. R-Style Softlab, part of Rosselkhozbank Group, has also introduced a solution capable of diagnosing crop diseases from images and recommending treatments. A comparable application, Agrolab, has been developed by students at Orenburg State University.
Care, Diagnosis and Treatment
This reflects a broader trend in Russia toward the development of applied AI models for agriculture. Looking ahead, one of the most promising directions is integrating AI diagnostics with robotic systems operating in fields, orchards or greenhouses. These may include both ground-based and aerial platforms capable of monitoring plant health, diagnosing diseases in real time and taking immediate action to treat affected crops and protect surrounding plants. In more complex cases, the system could automatically involve a human specialist.
Such robotic solutions are already being developed and continuously improved. Integration with digital phytopathology systems will enhance their effectiveness. Over time, these technologies are expected to become part of broader smart farming platforms being deployed by major agricultural producers.
Digitalization of plant care and disease management is expected to significantly increase food production in Russia. As domestic demand is met, a growing share of output could be directed toward export markets. In addition, digital platforms for automated plant diagnostics and treatment represent a potential export product. After adaptation, they could be deployed across CIS countries, the Middle East, North Africa and other fast-growing agricultural regions.
