Detection of Weed & Crop using YOLO v5 Algorithm

Abstract

In Agriculture, weeds are unwanted, as they compete with plants in absorbing essential elements such as nutrients, water, space & other growth requirements. This will affect the farmer by negatively influencing the yield & quality of crops. According to a study by ICAR, farmers spend huge sums of money to manage weeds. Also, weeds are ignored, since their impact is not visible unlike the visible impact of insects and diseases. Weed growth can have a devastating effect on the yield of crops. If weed growth is not stopped at critical moments, it will result in huge crop losses, sometimes up to 70%. Timely weeding is necessary to obtain an increased yield. A couple of weed removal approaches exist, of which the use of weeder machines is in practice, commonly. There have been advances in A.I & automation technology in recent times & the field of agriculture can also greatly benefit from these advances in technology. A.I Technology can be used to assist farmers in the process of weed management. Weed management can be monitored with the YOLO v5 model which is trained on a custom dataset of weeds, crops & infected leaves. The YOLO v5 model enables the feasibility of Multi-layer Neural Networks which can be used for the detection, recognition, & mapping of weeds during the initial growth stages of crops. This will contribute to a more effective, sustainable weed management approach.

Topics & Concepts