A cross-disciplinary student team has developed an innovative technology called AgriVision, an advanced prototype designed to detect diseases in rice plants using Artificial Intelligence (AI) and Internet of Things (IoT) technology. Developed through the Student Creativity Program in Creative Initiative (PKM-KC) under the supervision of Ir. Susilo Romadhon, S.Kom., M.Cs. from the Information Technology Study Program, the prototype was specifically designed to address bacterial leaf blight and blast disease, two threats that frequently affect rice crops in Indonesia.
The collaborative team is led by Muhammad Dzaki Pranaja Awish from Information Technology, together with Putraku Ruliff Abbas from Information Technology, who focuses on AI training and website development; Aryasetya Maulana and Muhammad Zaidaan from Electrical Engineering, who are responsible for electrical infrastructure, robotics, and telecommunications; and Tri Purwanti Ningsih from Biology Education, who plays a crucial role in scientifically validating parameters related to rice plant health.
“We actually started with a literature review. Rice is an important commodity in Indonesia because it is the country’s main source of carbohydrates. Bacterial leaf blight and blast are among the diseases most commonly found in rice crops in Indonesia. In practically every rice-growing area and every paddy field, there is a potential risk of these diseases,” said team leader Muhammad Dzaki Pranaja Awish.
To overcome limited internet connectivity in rice-growing areas, AgriVision relies on LoRa (Long Range) technology to transmit data locally to a receiver. The data can then be accessed directly by users through a mobile application or a localhost-based website without requiring conventional internet hosting services.
The prototype operates using a dual-detection system. A webcam scans the physical condition of rice plants from a distance of up to five meters using Computer Vision, while soil sensors simultaneously measure NPK nutrient levels and pH, as diseased rice plants tend to show indicators of potassium deficiency. The AI then classifies the data in real time, and when disease is detected, the system is designed to trigger an automated pesticide spraying command according to the required dosage.
“The most time-consuming parts of developing the device were the AI and the website. We had to ask ourselves whether there would actually be internet access in the rice fields, so we developed a website that uses LoRa technology. The AI was also challenging because we had to train it to detect bacterial leaf blight and blast disease,” Dzaki explained.
Following an intensive three-month development process, the team conducted four days of field testing across several rice fields in Boyolali, producing highly encouraging results. The AI system they trained demonstrated a high level of accuracy, making only two incorrect predictions out of 80 disease-classification trials. Although the device still faces challenges in rainy conditions, which can reduce the camera’s detection confidence, the team continues to refine the system. Going forward, they plan to expand its coverage so that it can scan an entire rice field plot from distances of up to eight meters by installing two camera poles diagonally across the field.
“Based on the four days of testing we carried out in the Boyolali area, we successfully classified bacterial leaf blight, blast disease, and healthy rice plants. The AI we developed was able to detect all of them with very few errors, making only two incorrect predictions out of 80 trials. However, rainy weather does affect the camera because its confidence level drops, and that remains one of our challenges alongside detection range,” Dzaki added.
The development of the hardware and AI system was supported by integrated funding of Rp8.7 million from Kemendikbudristek and the university’s internal funds. Given the high specifications of the electronic components used, around 75 percent of the budget was allocated solely to hardware assembly, while the remaining funds, supplemented by the team’s own contributions, were used to fulfill mandatory PKM outputs such as video production and field operational costs. In accordance with the core principles of PKM-Karsa Cipta, which focuses specifically on developing an initial prototype without involving community partners, AgriVision now serves as a promising foundation for precision agriculture technology that could be developed further to support the future welfare of the agricultural sector.



