Increasing Shape Bias to Improve the Precision of Center Pivot Irrigation System Detection

Irrigation is indispensable in agriculture. Center pivot irrigation systems are popular means of since they water-efficient and labor-saving. Monitoring center provides important information for the understanding agricultural production, water resources consumption environmental change. Deep learning has become an effective approach object detection semantic segmentation. Recent studies have shown that convolutional neural networks (CNNs) prone to be texture-biased rather than shape-biased, increasing shape bias can improve robustness performance CNNs. In this study, a simple yet method was proposed increase precision system detection. We extracted edge images training samples integrated them into data networks. With scheme, we evaluated compared PVANET YOLOv4. Experiments with Mato Grosso both YOLOv4 achieved improved performance, which demonstrated validity method.