Row Line Detection of Pineapple Planting Based on the RLYOLOv11 Model

This study integrates the RLYOLOv11 model with the RANSAC algorithm to identify the planting row lines of pineapples in a pineapple field, aiming to provide data support for precision agriculture. The study uses YOLOv11 as the baseline model and enhances small object detection by replacing the Conv in the YOLO neck with the attention-enhanced convolution RFAConv. Additionally, the original loss function was improved to the CIoU (Complete Intersection over Union) loss to enhance the model’s detection accuracy. After testing, the accuracy of the model in this study reached 84.4%, outperforming the R-CNN, YOLOv10, and DETR models. Furthermore, this study employs the RANSAC clustering algorithm to cluster the detected pineapple coordinates and performs least-squares fitting on different clusters. Experiments have shown that using the RLYOLOv11 model for detection followed by fitting can achieve the best results.