Related Experiment Video
Updated: May 8, 2026
05:16
Flying Insect Detection and Classification with Inexpensive Sensors
Published on: October 15, 2014
25.1K
Enhancing cotton whitefly (Bemisia tabaci) detection and counting with a cost-effective deep learning approach on the
Zhen Feng1, Nan Wang1, Ying Jin1
1The Key Laboratory for Quality Improvement of Agricultural Products of Zhejiang Province, College of Advanced Agricultural Sciences, Zhejiang A&F University, Linan, Hangzhou, 311300, Zhejiang, China.
Plant Methods
|October 19, 2024
Summary
This study developed an automated system using a deep learning model for fast cotton whitefly (Bemisia tabaci) detection and quantification. The system enables early pest identification, improving crop management strategies.
Area of Science:
- Agricultural Entomology
- Computer Vision
- Machine Learning
Background:
- Cotton whitefly (Bemisia tabaci) is a significant agricultural pest causing crop damage.
- Manual detection is labor-intensive and challenging due to pest characteristics.
- Need for efficient, high-throughput automated systems for real-time monitoring.
Purpose of the Study:
- Develop an automated system for efficient cotton whitefly detection and quantification.
- Enhance early pest detection for improved crop management.
- Create a tool for fast identification and counting of B. tabaci.
Main Methods:
- Developed a deep learning model based on YOLO v8s, modified with Swin-Transformer and a P2 structure.
- Compiled and augmented a dataset of 1200 annotated whitefly images.
- Integrated the model on a Raspberry Pi with a GUI and used SAHI for preprocessing.
Main Results:
- Achieved high performance with precision of 0.87, mAP50 of 0.92, and F1 score of 0.88.
- Demonstrated precise detection and counting of whiteflies on cotton leaves.
- Identified higher whitefly density in the afternoon and specific plant sections.
Conclusions:
- The enhanced YOLO v8s model enables precise whitefly detection and counting on hardware devices.
- The system is suitable for research requiring accurate quantification, such as phenotypic analysis.
- Future deployment in fields aims to manage whitefly infestations effectively.
