SegNeXt-RCMSCA: An improved SegNeXt network for detecting winter wheat lodging from UAS RGB images
Winter wheat lodging reduces wheat yield and poses a further risk to regional food security, emphasizing the need for timely and accurately monitoring of affected areas. Advances in Unmanned Aerial System (UAS) remote sensing and deep learning techniques provide new tools for detecting winter wheat...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-12-01
|
| Series: | Smart Agricultural Technology |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2772375525004617 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Winter wheat lodging reduces wheat yield and poses a further risk to regional food security, emphasizing the need for timely and accurately monitoring of affected areas. Advances in Unmanned Aerial System (UAS) remote sensing and deep learning techniques provide new tools for detecting winter wheat lodging. In this study, the RGB images of winter wheat were captured by the DJI Phantom 4 Pro V2.0 at flight heights of 60 m (0.8cm/pixel), 90 m (1.8cm/pixel), 120 m (3.1cm/pixel), and 150 m (4.3cm/pixel). A novel SegNeXt-RCMSCA network was proposed by integrating horizontal and vertical pooling with a multi-scale self-calibrated convolution function to enhance global contextual information. The SegNeXt-RCMSCA model achieved an Intersection over Union (IoU) of 86.72 %, F1 score of 92.89 %, Recall of 94.33 %, and Precision of 91.49 %. The model was tested using images with different spatial resolutions, and the results indicated that the 60 m (0.8cm/pixel) achieved the highest detection accuracy. The proposed SegNeXt-RCMSCA demonstrated strong potential for detecting lodging in other crops, offering a robust tool for improving the crop management in precision agriculture. By enabling timely and accurately lodging detection, the model facilitates crop damage assessment, harvest optimization, and informed field management, while supporting large-scale agricultural monitoring and intelligent decision-making in precision farming. |
|---|---|
| ISSN: | 2772-3755 |