Real-time detection of Chinese cabbage seedlings in the field based on YOLO11-CGB DOI Creative Commons
Hang Shi, Changxi Liu, Miao Wu

et al.

Frontiers in Plant Science, Journal Year: 2025, Volume and Issue: 16

Published: April 3, 2025

Accurate application of pesticides at the seedling stage is key to effective control Chinese cabbage pests and diseases, which necessitates rapid accurate detection seedlings. However, similarity between characteristics seedlings some weeds a great challenge for detection. This study introduces an enhanced method seedlings, employing modified version YOLO11n, termed YOLO11-CGB. The YOLO11n framework has been augmented by integrating Convolutional Attention Module (CBAM) into its backbone network. module focuses on distinctive features Additionally, simplified Bidirectional Feature Pyramid Network (BiFPN) incorporated neck network bolster feature fusion efficiency. synergy CBAM BiFPN markedly elevates model's accuracy in identifying particularly distant subjects wide-angle imagery. To mitigate increased computational load from these enhancements, network's convolution replaced with more efficient GhostConv. change, conjunction network, effectively reduces size requirements. outputs are visualized using heat map, Average Temperature Weight (ATW) metric introduced quantify map's effectiveness. Comparative analysis reveals that YOLO11-CGB outperforms established object models like Faster R-CNN, YOLOv4, YOLOv5, YOLOv8 original YOLO11 detecting across varied heights, angles, complex settings. model achieves precision, recall, mean Precision 94.7%, 93.0%, 97.0%, respectively, significantly reducing false negatives positives. With file 3.2 MB, 4.1 GFLOPs, frame rate 143 FPS, designed meet operational demands edge devices, offering robust solution precision spraying technology agriculture.

Language: Английский

Real-time detection of Chinese cabbage seedlings in the field based on YOLO11-CGB DOI Creative Commons
Hang Shi, Changxi Liu, Miao Wu

et al.

Frontiers in Plant Science, Journal Year: 2025, Volume and Issue: 16

Published: April 3, 2025

Accurate application of pesticides at the seedling stage is key to effective control Chinese cabbage pests and diseases, which necessitates rapid accurate detection seedlings. However, similarity between characteristics seedlings some weeds a great challenge for detection. This study introduces an enhanced method seedlings, employing modified version YOLO11n, termed YOLO11-CGB. The YOLO11n framework has been augmented by integrating Convolutional Attention Module (CBAM) into its backbone network. module focuses on distinctive features Additionally, simplified Bidirectional Feature Pyramid Network (BiFPN) incorporated neck network bolster feature fusion efficiency. synergy CBAM BiFPN markedly elevates model's accuracy in identifying particularly distant subjects wide-angle imagery. To mitigate increased computational load from these enhancements, network's convolution replaced with more efficient GhostConv. change, conjunction network, effectively reduces size requirements. outputs are visualized using heat map, Average Temperature Weight (ATW) metric introduced quantify map's effectiveness. Comparative analysis reveals that YOLO11-CGB outperforms established object models like Faster R-CNN, YOLOv4, YOLOv5, YOLOv8 original YOLO11 detecting across varied heights, angles, complex settings. model achieves precision, recall, mean Precision 94.7%, 93.0%, 97.0%, respectively, significantly reducing false negatives positives. With file 3.2 MB, 4.1 GFLOPs, frame rate 143 FPS, designed meet operational demands edge devices, offering robust solution precision spraying technology agriculture.

Language: Английский

Citations

0