Real-time detection and classification of plant seeds using YOLOv8 object detection model

Abstract

This study addresses the pressing need for efficient and accurate plant seed detection and classification in agriculture by leveraging deep learning and computer vision technologies. Traditional manual methods for seed identification are time-consuming and labor-intensive, prompting the adoption of deep learning-based object detection techniques. This study focuses on implementing a real-time plant seed detection and classification system using the You Only Look Once (YOLO) v8 object detection model. The project workflow involves collecting a custom dataset of high-resolution images of various plant seeds, annotating them with bounding boxes, and preprocessing them to ensure uniformity and increase variability through data augmentation. The dataset is then partitioned into training, testing, and validation sets for model training and evaluation. The YOLOv8 model is trained on the annotated dataset, and its performance is rigorously evaluated. Additionally, related works in the field of seed classification and quality testing using deep learning techniques are reviewed, providing valuable insights into the application of such methodologies in agricultural practices. Experimental results demonstrate the effectiveness of the proposed approach, with the YOLOv8 model achieving high precision and recall rates. The deployment of the trained model on the Roboflow platform enables real-time seed detection and classification, showcasing its potential for practical agricultural applications. Overall, this study contributes to the advancement of agricultural automation and precision farming, offering a scalable and efficient solution for plant seed detection and classification that holds significant implications for improving crop management strategies.