Politeknik Negeri Malang
Rice farming, a cornerstone of Indonesia's economy, faces significant challenges from plant diseases that can cause substantial crop loss. Farmers often lack early detection methods, leading to delayed treatment and reduced yields. This highlights the urgent need for an accessible and efficient solution to help farmers quickly identify and manage rice leaf diseases.
Disoriza utilizes advanced image processing and YOLO-based object detection to identify diseases in rice plants accurately. The application processes images captured by the user, analyzes them using a trained machine learning model, and instantly provides comprehensive information about the disease, including its type, symptoms, and effective treatment methods.
User simply captures or uploads a photo of a rice leaf suspected of having a disease. The application instantly analyzes the image, detects the presence of disease, identifies the specific type of disease, and provides comprehensive information about symptoms and recommended treatment methods.
One of the main challenges was optimizing the machine learning model to run efficiently on mobile devices while maintaining high accuracy. This involved experimenting with different YOLO model variations and fine-tuning the training process to ensure fast inference times without compromising detection quality.