AI-Based Automated Weed Detection System Using YOLOv5 and Mask R-CNN for Sustainable Agriculture

Weeds are a big problem for farming, leading to lower crop yields and more use of herbicides and chemicals, affecting economic and environmental sustainability. AI algorithms, especially deep learning techniques, have become a hopeful option in the field of smart precision agriculture. Utilizing advances in machine learning and deep learning, automated weed detection systems can greatly improve agricultural efficiency and reduce environmental damage. This study aims to tackle the spread of weeds in farm fields by creating a strong detection system powered by Artificial Intelligence. We start by conducting a thorough examination of the different kinds of weeds, their occurrence, and how they affect agriculture, laying the groundwork for comprehending the issue. An in-depth examination of the most recent DL techniques and uses in weed identification is carried out, focusing on the advantages and drawbacks of current methods. In order to improve existing abilities, we suggest implementing an automated weed identification system that incorporates two cutting-edge DL algorithms: YOLOv5 and Mask R-CNN. These models are used with a specialized Weed Dataset gathered in Tébessa province, made to mirror the distinct features of nearby farmlands. Even though the dataset is small, the experimental outcomes show promising performance, showcasing the potential of the suggested system in practical uses. This research helps advance smart farming by allowing accurate, effective, and focused weed control. The system helps to decrease the amount of chemicals and manual work needed, leading to the adoption of more eco-friendly farming methods. Future studies will concentrate on enhancing the dataset, improving model efficiency, and incorporating the system into complete agricultural workflows.