Advancing Weed Detection in Agricultural Landscapes Using Computer Vision

This paper explores the need for weed detection in agricultural landscapes, where weed infestations pose significant crop productivity challenges. To address this matter, computer vision techniques are utilised to create a weed detection algorithm that accurately detects weed patches inside agricultural field images. To ensure reliable model evaluation, the dataset is split into training and testing subsets before being fed into the YOLOv3 model. The YOLOv3 model performs well in terms of precision, with an average precision of 66.4
^%
for crop regions and a notable precision of 92.5% for detecting weeds. These results highlight the potential of artificial intelligence-driven weed detection techniques to optimize crop management practices and foster sustainable agriculture in the modern era. This research advances precision agriculture by effectively addressing weed infestations, paving the way for more efficient and targeted weed control strategies, which ultimately lead to increased crop yield and improved agricultural sustainability.