This is the authors' abstract. We don't add key points for this paper.
The current scoop-chain potato seed metering device has the problems of high replanting rate, high seed leakage rate and poor seeding performance. In order to quickly and accurately identify the seed blocks on the conveying bucket chain of potato, a detection model of YOLOv8n-CSQG algorithm was proposed. The model replaced C2f with CSPPC to improve the reasoning speed; To enhance occlusion image recognition, the feature fusion layer is replaced with Slim-Neck and integrated with GSCongv; VoVGSCSP, grouped convolution, and channel rearrangement were used to construct a lightweight detection head. In addition, QFL and GIoU loss functions are introduced to optimize speed and accuracy. The experimental results show that compared with the baseline model, the detection accuracy of YO8n-CSQG is increased by 10.21%, the recall rate is increased by 6.3%, the frame rate (FPS) is increased by 45.25%, the calculation amount (FLOPs) is reduced by 25.92%, the parameter amount is only 2.3 million, and the system is lightweight and efficient. The model can accurately identify the missing pieces and multiple pieces in the spoon-chain potato picking device, and provide technical support for the control and model deployment of the terminal executor of the follow-up intelligent replanting robot.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Sun et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: