The research team at TMA Solutions developed the T-Pest system integrating AI, IoT and geographic information system (GIS) to identify 7 types of diseases and warn of rice pests.
The product is made from the topic “Research on the application of artificial intelligence in identifying and warning some rice pests and diseases in Binh Dinh province”. T-Pest system, including an automatic insect data collector and a phone application to identify diseases on rice plants.
According to engineer Tran Hoan Anh Nguyen, Director of Data Science Center, Tuong Minh Software Solutions Co., Ltd. (TMA Solutions), with a large cultivated area and the specific climate of the Central region, Binh Dinh province often facing outbreaks of rice pests and diseases. Previously, farmers mainly relied on experience and manual methods to identify and handle them, which were both time-consuming and inaccurate. Using information from meteorology to combine and make predictions about the possibility of pests and diseases is not effective compared to placing equipment directly in the field.
T-Pest machine is located in Phuoc Son commune, Tuy Phuoc district. Image: Thao Chi
The T-Pest system operates automatically, including high-resolution cameras, UV lights and environmental sensors. Device attracts insects and records images. All collected data is analyzed by an innovative AI model based on YOLOv5-Ghost, helping to accurately identify and classify pests.
The highlight of the system is its ability to operate automatically 24/7, helping to promptly record and warn about pest situations. All data is integrated on a web management platform, allowing real-time monitoring of pest developments and visual display on thematic maps.
T-Pest currently recognizes 7 types of diseases (leaf blight, rice blast, brown spot, physiological leaf yellowing, sheath blight, flat grain, bacterial streak spot) and 8 types of insects (white-backed planthopper, brown planthopper, caterpillar). stem borers, small leaf rollers, onion gnats, black-tailed green leafhoppers, black stink bugs, green stink bugs).
Insect images were collected from the T-Pest machine. Image: Thao Chi
According to engineer Nguyen, to implement the project, the research team built a detailed database of common pests and diseases in Binh Dinh fields. Data is collected through photos taken from smartphones and specialized computer systems. All images are collected from rice fields, then labeled with the correct disease type, confirmed by experts.
Based on the database, the team designed the T-Pest system. In addition to the function of attracting, identifying and automatically classifying rice diseases, the T-Pest system also has the ability to collect and present data on cultivated area, rice variety information, season and pest information. disease in rice. At the same time, provide response solutions from management agencies; Thematic map of equipment management and pest and disease status in rice throughout the region. All collected data will be sent to the server and managed on the website, allowing users to track data, receive pest forecasts and manage information conveniently.
Mr. Nguyen Quoc Duong, a member of the research team, said that improving the YOLOv5 model is an important breakthrough in research to develop a solution capable of quickly and accurately identifying diseases and harmful insects in rice. body. This helps streamline the model, minimizing the space and resources needed while still maintaining high performance. This innovative architecture not only improves accuracy but also ensures flexible deployment under real field conditions, playing a core role in automating pest monitoring.
Through testing at 5 locations in 3 communes: Cat Nhon (Phu Cat), Phuoc Son and Phuoc Thuan (Tuy Phuoc), the system has collected more than 1,000 insect images. The results show that the accuracy in identifying dangerous insects and pests such as black stink bugs, green stink bugs, and brown planthoppers reaches over 80% in real conditions. In addition, the disease identification application on mobile phones also helps farmers easily look up information, receive early warnings and apply timely treatment solutions.
Mr. Nguyen Van Chin, farmer in Phuoc Son commune, Tuy Phuoc district shared: This T-Pest machine helps farmers monitor rice fields and detect pests easily without having to go to the fields regularly. I would feel more secure if I had a reliable tool to help me take care of my plants and crops.
Authorities tested the T-Pest machine’s ability to operate when actually deployed in the field. Image: Thao Chi
MSc Le Hoai Lam, Deputy Director, Binh Dinh Department of Crop Production and Plant Protection highly appreciated the features and efficiency of the system. “Recently, we have used this system to refer to information, helping to determine the growth stage of pests and diseases, thereby taking timely treatment measures, contributing to minimizing losses for farmers.” .
However, according to Mr. Lam, the system has only been tested in 3 communes and can only identify a few types of insects. For more effective application, it is necessary to add more data on pests and expand the deployment area of T-Pest machines.
According to engineer Nguyen, after acceptance, TMA Solutions will transfer technology and support the expansion of this model in many other areas of Binh Dinh province, contributing to modernizing pest control in the province. local agricultural production. With the results achieved, the research promises to open up new directions for agriculture 4.0 in Binh Dinh.