A Real-Time YOLOV9-Based Approach for Precision Weed Detection in Peanut Fields

Even though a wide variety of weeding machines are available for both small and large-scale farming, the automatic detection of weeds from crops remains a significant challenge in modern agriculture, critically affecting crop yields, sustainability, and operational efficiency. This work proposes a smart weed detection system that identifies the real time peanut plant from the weeds and yields higher production. The YOLOv8 and YOLOv9 object detection is used to identify the peanut plant and weeds precisely. The real time image is acquired by an ESP32 camera module placed above the weed machines. The Wireless vision capabilities are made possible by the integration of the ESP32 camera, which takes highresolution field photos in a variety of settings and sends them in real time to the YOLOv9-S detection with processing speed of 0.6 second per image. The captured image has been preprocessed with data annotation and augmentation for model processing. The YOLOv9-M model demonstrated superior performance, achieving a precision of 89.4 %. This work effectively distinguishes weeds from peanut plant under diverse and challenging field conditions, such as dense vegetation, variable lighting, and overlapping plant structures. The results establish YOLOv9-M integrated with ESP32 camera promised a significant precision agriculture as a real-time tool, providing efficient and sustainable weed control management.