Precision agriculture (PA) is a technology that integrates AI into the agricultural field. One of the uses of PA is to increase the productivity of rice plants, which is a staple food in high demand. In this study, PA technology based on vision computing will be developed, where the multiclass intelligent object detection model will be embedded into edge computing devices. The proposed model, which is called YOLO-Rice, is an improved version of You Only Look Once version 8 (YOLOv8), in which attention modules including GhostNet and the convolutional block attention module (CBAM) are combined to enhance the detection performance and model efficiency. There are 12 classes divided into three categories: rice grain, rice leaves, and insect pests. The stages of model development are carried out through dataset development, model construction, model evaluation and validation, and model deployment. The comparative experiment incorporates YOLOv5, YOLOv8, YOLOv8+CBAM, YOLOv8+GhostNet, and YOLOv9. The mean average precision value (mAP50) of YOLO-Rice was the most optimal compared with other deep learning algorithms, attaining an accuracy percentage of 91.95%. Overall, based on the experimental results, YOLO-Rice has outstanding results in terms of detection accuracy and model efficiency.