Lightweight Object Detection Model for False Smut Detection in Rice Leaf with Eigen-Cam Interpretability

Agriculture plays a vital role in many economies, with rice being a key staple crop; however, diseases such as false smut significantly affect yield and quality. This paper proposes a cost-effective and real-time disease detection system using a lightweight YOLOv8 model deployed on a Raspberry Pi. The system captures rice leaf images through a camera module and accurately identifies infected regions. Additionally, IoT-based sensors are used to monitor soil moisture and pH levels, and the collected data is transmitted to a cloud platform for analysis. Based on detection results and environmental conditions, automated actions such as activating a water pump or sprayer are performed using a relay module. The proposed system reduces manual effort, enables early disease detection, and improves overall crop productivity.