Related Experiment Video
Updated: Jun 5, 2025
11:49
Cereal Crop Ear Counting in Field Conditions Using Zenithal RGB Images
Published on: February 2, 2019
9.2K
YOLO-SDL: a lightweight wheat grain detection technology based on an improved YOLOv8n model
Zhaomei Qiu1, Fei Wang1, Weili Wang1
1College of Agricultural Equipment Engineering, Henan University of Science and Technology, Luoyang, Henan, China.
Frontiers in Plant Science
|December 5, 2024
Summary
A new deep learning model, YOLO-SDL, efficiently detects wheat grain quality using advanced architecture. This AI solution enhances agricultural automation with high accuracy and speed, even in limited environments.
Area of Science:
- Agricultural Science
- Computer Vision
- Artificial Intelligence
Background:
- Wheat grain quality is vital for food safety and economic stability.
- Traditional wheat grain detection methods are inefficient.
- Deep learning offers advanced solutions for accurate grain recognition.
Purpose of the Study:
- To develop an improved, efficient deep learning model (YOLO-SDL) for wheat grain detection.
- To enhance detection speed and accuracy while maintaining a lightweight model.
- To provide a technical solution for agricultural automation.
Main Methods:
- Constructed a high-quality wheat grain dataset with diverse grain conditions.
- Applied data augmentation techniques to increase dataset complexity.
- Developed the YOLO-SDL model using ShuffleNetV2 backbone and DWConv with LSKA mechanism in the neck.
Main Results:
- YOLO-SDL achieved superior performance in wheat grain detection.
- The model demonstrated high precision (P=0.942) and recall (R=0.903).
- Achieved mAP50 of 0.965 and mAP50-95 of 0.859 with low computational complexity.
Conclusions:
- The YOLO-SDL model offers an efficient balance of lightweight design and performance optimization.
- ShuffleNetV2, DWConv, and LSKA structures proved effective for wheat grain detection.
- The model serves as a reliable reference for agricultural automation and crop detection.
