Improving Wheat Spike Counting and Disease Detection with Deep Learning
Author Information
Author(s): Li Ruiheng, Hong Wenjie, Wu Ruiming, Wang Yan, Wu Xiaohan, Shi Zhongtian, Xu Yifei, Han Zixu, Lv Chunli, Peteinatos Gerassimos, Moreno Hugo
Primary Institution: China Agricultural University, Beijing, China
Hypothesis
This study aims to improve the precision of wheat spike counting and disease detection using deep learning techniques.
Conclusion
The proposed model significantly enhances the accuracy and efficiency of wheat spike counting and disease detection in precision agriculture.
Supporting Evidence
- The model achieved a precision of 0.93, recall of 0.89, accuracy of 0.91, and mAP of 0.90 in disease detection.
- In wheat spike counting, the model demonstrated a precision of 0.91, recall of 0.88, accuracy of 0.90, and mAP of 0.90.
- The proposed model outperformed existing models like YOLOv8 and WheatNet in both tasks.
- A density loss function was introduced to improve model performance in dense regions.
- Comparative experiments showed significant improvements in detection accuracy with the proposed methods.
Takeaway
This study shows how computers can help farmers count wheat spikes and find diseases in plants more accurately and quickly.
Methodology
The study used deep learning models with a probability density attention mechanism to analyze images of wheat spikes and diseases.
Limitations
The study may be limited by the quality and diversity of the image dataset used for training the models.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website