Designing and Developing a Weed Detection Model for California Thistle

With a great percentage of farms in New Zealand as pastures, they are mainly important contributing to the milk and meat industries. Pasture quality is highly affected by weeds. Weeds grow fast invade pastures seed pollination. They consume nutrients, water, other minerals, once bitter, cattle do not eat them. Therefore, dairy farmers have allocate significant portion their budget time monitor clean Unfortunately, most weed management tasks manual with no consistent technology. Thus, motivation behind this paper was design an object detection model for monitoring control pastures. The designed tested on California thistle, dominant widespread Our study one major designs identifying weeds in-pasture environment, complicated environments any model. A synthetic methodology used create three types datasets: plant-based, leaf-based, mixed. trained based leaf-based dataset contributions our work which has been conducted models. After models had trained, tuning experimentation undertaken improve model’s performance. This involved studying hyperparameters various ranges then recording values at optimum points. improved showed 93% mAP accuracy training images over 95% testing images. that slightly better than can automate system. use will save money help them reduce errors work.