Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Abstract. Grape downy mildew (GDM) causes serious damage to grape production, yield and quality. GDM disease severity classification is one of the important prerequisites for precise variable pesticide application in grape orchards. Unlike traditional manual disease detection relies on farm experts and is often time-consuming, computer vision and artificial intelligence technology provide automatic disease grading for real-time controlling the spread of disease on the grapevine. To improve GDM severity classification under natural environments, a deep learning based approach SE-ResNet18 is proposed in this study. The proposed SE-ResNet18 combing ResNet18 and SE mechanism to enhance GDM disease feature representation ability and severity classification performance. A GDM dataset was acquired in a vineyard under a nature scene (consisting of different illuminations, shadows and backgrounds) to test the proposed approach. Experimental results show that the proposed SE-ResNet18 based approach achieved a GDM severity classification accuracy of 96.25%, higher than that of VGG16, and ResNet18. The proposed method in this study is favorable for the classification of GDM severity level, which is helpful in the automatic plant leaf disease control in digital vineyard farming.