A rapid and precise algorithm for maize leaf disease detection based on YOLO MSM
These videos have been matched automatically. Contact us if they are not relevant.
These videos have been matched automatically. Contact us if they are not relevant.
View abstract on PubMed
Summary
This summary is machine-generated.A new YOLO-MSM algorithm improves maize leaf disease detection using multi-scale variable kernel convolution and attention mechanisms. This lightweight model achieves high accuracy and speed, enabling mobile device deployment for early disease identification.
Area Of Science
- Agricultural Science
- Computer Vision
- Machine Learning
Background
- Accurate and real-time maize leaf disease detection is crucial for reducing agricultural economic losses.
- Challenges in current methods include large datasets, low accuracy, and production inefficiencies.
Purpose Of The Study
- To introduce YOLO-MSM, an advanced algorithm for maize leaf disease detection.
- To enhance detection accuracy, speed, and efficiency in real-world agricultural settings.
Main Methods
- Developed the MKConv (Multi-scale Variable Kernel Convolution) for adaptive feature extraction.
- Integrated the C2f-SK module with Selective Kernel (SK) attention for optimized feature representation.
- Utilized MPDIoU (Minimum Point Distance Intersection over Union) loss function for improved target localization.
Main Results
- YOLO-MSM achieved a real-time detection rate of 279.56 frames per second (fps).
- Demonstrated improvements in precision (0.66%) and recall (1.61%) compared to baseline algorithms.
- The algorithm is lightweight (5.4 MB) with significantly reduced parameters and FLOPs.
Conclusions
- YOLO-MSM offers a superior balance between precision and speed for maize leaf disease detection.
- The lightweight design facilitates deployment on mobile devices for practical agricultural applications.
These videos have been matched automatically. Contact us if they are not relevant.
These videos have been matched automatically. Contact us if they are not relevant.
