Abstract: It is crucial for agricultural production that weeds be reduced. Because they compete with crops for essential resources like water, nutrients, and sunlight, weeds can severely reduce crop yields. It is essential to choose the appropriate herbicides, choose planting locations, and manage cultivation depth to reduce crop damage while managing weeds efficiently. The goal of the project is to create a deep learning-based system for automatically identifying weeds in agricultural fields to overcome this constraint. Convolutional neural networks (CNNs) are used in this proposed system since they are effective for picture categorization tasks. A big dataset named Weed, which consists of high-resolution images of numerous weed species, is put together to train and test the systemYOLOv3, YOLOv5, and Faster R-CNN are among the deep learning detection models that were trained using the Weed dataset. This shows that the Weed dataset may someday prove to be a useful training resource for the creation of a real-time weed identification program.

 
Keywords: Weed Identification, Deep Learning, CNN Algorithm, Image processing, Color Index.

| DOI: 10.17148/IARJSET.2023.10713