An Adaptive Row Alignment Method for Sea Buckthorn Pruning Harvester Based on YOLOv8 Model

Sea buckthorn is a shrub with both ecological and economic values. Mechanized harvesting of sea buckthorn significantly enhances harvesting efficiency; however, the cab of the sea buckthorn pruning harvester is located at the side, creating a blind spot. This blind spot prevents the operator from accurately manipulating the pruning table to achieve row alignment with sea buckthorn trees. To address this issue, we propose an adaptive row alignment method for the sea buckthorn pruning harvester based on the YOLOv8 model, which aims to provide robust technical support for automated harvesting. The method begins by capturing images via a camera mounted on the harvester, followed by preprocessing the sea buckthorn fruit images through grayscale conversion and binarization. The YOLOv8 model is then employed to identify sea buckthorn fruits in the binarized images. Finally, the centerline of the fruit regions is fitted using the leastsquares method to generate the navigation line for the sea buckthorn pruning harvester during operation.