Lightweight Model for Detecting Grape Leaf Diseases
Author Information
Author(s): Fu Zilong, Yin Lifeng, Cui Can, Wang Yi
Primary Institution: College of Rail Intelligent Engineering, Dalian Jiaotong University, Dalian, China
Hypothesis
Can a lightweight MHDI-DETR model improve the detection of grape leaf diseases while reducing computational requirements?
Conclusion
The MHDI-DETR model significantly reduces parameters and computational load while improving detection accuracy for grape leaf diseases.
Supporting Evidence
- The MHDI-DETR model achieved a 56% reduction in parameters and a 49% reduction in floating-point operations compared to the RT-DETR model.
- The model achieved precision rates of 96.9%, 92.6%, and 72.5% for accuracy, mAP50, and mAP50:95, respectively.
- The integration of GIoU and Focaler-IoU into Focaler-GIoU enhanced detection accuracy and convergence speed for small targets.
Takeaway
This study created a smart model that helps farmers quickly find sick grape leaves without needing a lot of computer power.
Methodology
The study used a lightweight model architecture combining MobileNetv4 and various advanced techniques to enhance detection accuracy and efficiency.
Limitations
The model's performance may vary under different environmental conditions and with different datasets.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website