Related Experiment Video
Updated: Jul 28, 2025
08:47
Author Spotlight: UAV Remote Sensing for Efficient Invasive Plant Biomass Estimation
Published on: February 9, 2024
1.5K
Crop pest detection by three-scale convolutional neural network with attention
Xuqi Wang1, Shanwen Zhang1, Xianfeng Wang1
1College of Information Engineering, Xijing University, Xi'An, 710123, China.
Plos One
|June 2, 2023
Summary
A new three-scale convolutional neural network with attention (TSCNNA) model improves crop pest detection accuracy. This AI approach enhances identification of small pests in complex field conditions for better agricultural security.
Area of Science:
- Agricultural Science
- Computer Vision
- Artificial Intelligence
Background:
- Crop pests significantly impact agricultural yield and quality, necessitating accurate detection for effective management.
- Existing convolutional neural network (CNN) methods struggle with recognizing small, diverse pests in complex field environments.
Purpose of the Study:
- To develop an advanced CNN model for improved crop pest detection, particularly for small and varied pest species.
- To enhance the accuracy and efficiency of real-time pest identification in agricultural fields.
Main Methods:
- A novel three-scale CNN with attention (TSCNNA) model was developed, incorporating channel attention and spatial mechanisms.
- The model was designed to improve CNN's focus on pests of varying sizes against complicated backgrounds and enlarge its receptive field.
Main Results:
- The TSCNNA model achieved a precision of 93.16% in crop pest detection.
- This represents a significant improvement of 5.1% and 3.7% over existing ICNN and VGG16 methods, respectively.
Conclusions:
- The TSCNNA model demonstrates high speed and accuracy for crop pest detection.
- This method holds practical significance for real-time, in-field pest control and safeguarding crop production.
