基于YOLOV5的柑橘果实目标检测研究

Research on Identification and Detection of Citrus Fruit Based on YOLOV5

为研究自然环境下自主采摘机器人对柑橘果实的快速识别技术,实现柑橘的早期产量预测,提出一种基于YOLOv5算法的柑橘果实识别方法。该方法通过Mosaic增强图像数据,在YOLOv5的基础上修改回归框损失函数,并优化相关网络参数加速模型收敛,加强对小目标的检测。实验结果表明,改进的YOLOv5算法显著提升了召回率和输出预测框的检测精度。

In order to study the rapid recognition technology of citrus fruit by autonomous picking robot in natural environment and realize the early yield prediction of citrus, a citrus fruit recognition method based on yolov5 algorithm is proposed. This method enhances the image data through mosaic, modifies the regression box loss function on the basis of yolov5, and optimizes the relevant network parameters to accelerate the model convergence and strengthen the detection of small targets. The experimental results show that the improved yolov5 algorithm significantly improves the recall rate and the detection accuracy of the output prediction frame.