Crop-Weed Segmentation and Classification Using YOLOv8 Approach for Smart Farming

Accurately segmenting crop and weed images in agricultural fields is crucial for precision farming and effective weed management. This study introduces a new method that leverages the YOLOv8 object detection model for precise crop and weed segmentation in challenging agricultural scenes. Our approach involves preprocessing agricultural images to enhance feature representation, followed by YOLOv8 for initial crop and weed detection. Thorough experiments using standard datasets comprising 2630 images demonstrate the effectiveness of our proposed method concerning precision, recall, mean average precision (mAP), and F1 score compared to existing techniques. These findings contribute to advancing crop-weed segmentation techniques, offering practical solutions for efficient weed management and precision agriculture. Our proposed approach outperforms state-of-the-art methods found in the literature. Our methodology presents a promising framework for automated crop-weed segmentation with applications in crop monitoring, yield estimation, and weed control strategies, supporting sustainable agricultural practices.