Authors

Abstract

In order to realize the rapid and accurate detection of wheat impurities, this study proposes a wheat impurity detection algorithm based on improved YOLO v8. This method significantly reduces the number of model parameters by improving the original C2f module to C2f_UIB module. In addition, High-level Screening-feature Fusion Pyramid Networks ( HS-FPN ) is introduced to solve the problem of scale difference between straw and wheat ear. Finally, GIoU is introduced to deal with the scene with high density impurities. This study combines the UIB module of MobileNetV4 with HS-FPN for the first time, and proposes a triple optimization framework for agricultural lightweight detection.

Abstract in Chinese

为实现对小麦杂质的快速准确的检测，本研究提出一种基于改进YOLO v8的小麦杂质检测算法。该方法通过改进原有C2f模块为C2f_UIB模块，显著降低了模型参数量。此外，引入高级筛选特征融合金字塔网络（HS-FPN），用于解决秸秆和麦穗这两类杂质的尺度差异的问题。最后通过引入 GIoU 来应对有着高密度杂质的场景。本研究首次将MobileNetV4的UIB模块与HS-FPN结合，提出面向农业轻量化检测的三重优化框架。