<![CDATA[The suitability of rice varieties to agroclimatic conditions is a key factor in determining rice productivity in Indonesia. Climate variability and land limitations require a decision support system capable of assisting farmers in selecting rice varieties suitable for local environmental conditions. This study aims to develop an integrated artificial intelligence-based system that combines YOLOv9 for image-based rice variety recognition and Naïve Bayes for climate suitability prediction based on temperature and humidity parameters. Image data of five rice varieties Ciherang, Inpari 32, Inpari Nutrizinc, Mekongga, and Baroma were collected directly from agricultural fields in Bandung Regency and processed through annotation, augmentation, and model training stages. The YOLOv9 model performed well in distinguishing rice varieties with relatively similar morphological characteristics, with an mAP@50 value of 0.8932. Meanwhile, the Naïve Bayes model achieved 78% accuracy in predicting climate suitability based on altitude, temperature, and humidity, and produced predictions consistent with agronomic recommendations. Both models were then integrated into a Gradio-based interactive interface to facilitate use by non-technical users. The results indicate that this integrated approach has the potential to be an effective decision support system for assisting in the selection of rice varieties that are adaptive to microclimate conditions, thereby supporting more efficient and sustainable rice cultivation practices.]]>
Copyrights © 2026
