Agricultural techniques have been transformed by integrating robots and computer vision, especially in seedling transplanting and detection automation. This work presents a novel method for real-time seedling identification in diverse environmental conditions utilizing the YOLOv5 (You Only Look Once version 5) deep learning model. After training on a large dataset of images of seedlings, the model demonstrated excellent classification and localization accuracy. An autonomous robotic device equipped with a robotic arm and camera was developed to expedite the transplant procedure. To facilitate accurate and effective transplanting, the method utilizes YOLOv5 to recognize seedlings and determine the optimal locations for harvesting. The experimental findings demonstrate significant improvements in accuracy and speed compared to conventional approaches. It reduces the risk of harm to seedlings and increases operating efficiency while minimizing the need for human labor. The results highlight the benefits of incorporating advanced AI and robotics into smart agriculture, as well as new methods for more sustainable agricultural techniques. © 2025 Elsevier B.V., All rights reserved.