Detection of weeds by image processing method in maize fields

Maize is one of the plants of great agricultural importance. One of the most important cereals in the world is maize. However, maize plant is a sensitive plant in terms of weed control. These weeds negatively affect the growth and yield of the maize plant. Therefore, weed detection and control is of great importance in maize cultivation. It is aimed to use image processing to identify these weeds up to the v5 development stage of maize, providing great convenience to agricultural producers and providing automatic detection and control. These technologies are important tools to increase agricultural yield and enable weed control. In this study, we created our own image datasets using machine learning with Roboflow and created our own dataset by separating and detecting weeds from corn in the images to be used in training. The dataset we created online via Roboflow was added to Google Colab. By including the Yolo v9 deep learning library, which is one of the pioneers of high accuracy object identification algorithms, it is aimed to detect weeds other than corn on the data set we have created with the algorithm we have developed. In addition, the parts where weeds are dense in the corn field were determined as regions. Thus, by spraying those areas locally, it is ensured to save both the cost of medication and to make a more environmentally friendly system at the same time since healthy corns will be exposed to less chemicals.