Weed Identification in Subterranean Vegetables using YOLO and Flask Framework

Effective weed identification in subterranean vegetable lands is essential for sustainable agriculture, as weeds compete with crops for nutrients, water, and light. Traditional weed management methods are labor-intensive and often involve excessive use of herbicides, which can negatively impact soil health and crop quality. This work introduces a precisiondriven approach to weed detection using advanced deep learning models YOLOv5 and YOLOv8. The models were trained on a dataset specifically designed to address the unique challenges of subterranean crop environments, where weeds often blend with soil and vegetation textures. YOLOv5 achieved a precision rate of 96.7%, while YOLOv8 further enhanced detection accuracy, reaching a precision rate of 98.2%. These high-precision rates demonstrate the effectiveness of YOLO architectures in distinguishing weeds from crops in complex agricultural settings. The system can be integrated into automated agricultural machinery for real-time weed identification and removal, reducing manual labour requirements and minimizing chemical herbicide use. This approach supports environmentally friendly farming practices by promoting weed control solutions that protect soil biodiversity and crop health. The results of this work highlight the potential of YOLO-based models as valuable tools in advancing precision agriculture in subterranean vegetable production systems.