Smart sprayer for weed control using an object detection algorithm (yolov5)

Abstract

Spraying pesticides is one of the most common procedures that is conducted to control pests. However, excessive use of these chemicals inversely affects the surrounding environments including the soil, plants, animals, and the operator itself. Therefore, researchers have been encouraged to develop robotic sprayers that can apply pesticides at variable rates as needed in the field. In this study, a remotely controlled sprayer with two modes (variable rate and constant rate applications) was developed and evaluated for some spray characteristics and application accuracy metrics when controlling weeds at two travel speeds. The variable rate mode resulted in a high precision, recall, and accuracy in detecting weed and applying herbicide that was 90%, 100%, and 94%, respectively. Moreover, the spray coverage, droplet density, and the deposition on weed using the variable rate mode were 34.16%, 127.64 deposites ∙ cm–2, and 7.67 μl ∙ cm–2, respectively. The result also revealed that the spray coverage, droplet density, and the deposition were less sensitive to the travel speed when adopting the variable rate mode compared to the constant rate mode.