Related Experiment Video
Updated: Apr 23, 2026
03:31
Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
Published on: December 15, 2023
1.3K
Intelligent Pest Recognition Based on Improved Convolutional Neural Networks with Multi-scale Hybrid Attention
1School of Electronics and Information Engineering, Hangzhou Dianzi Univ, Hangzhou, China. nijiangong@hdu.edu.cn.
Neotropical Entomology
|April 21, 2026
Summary
An intelligent pest recognition system using convolutional neural networks (CNNs) improves agricultural pest identification accuracy. This deep learning approach offers a more efficient and precise solution for crop protection and management.
Area of Science:
- Agricultural Science
- Computer Science
- Artificial Intelligence
Background:
- Pest damage significantly impacts agricultural yield and quality.
- Manual pest identification is inefficient, subjective, and resource-intensive.
- Advanced automated methods are needed for accurate pest detection.
Purpose of the Study:
- To develop an intelligent pest recognition system using deep learning.
- To enhance pest classification accuracy and efficiency in agriculture.
- To provide a technological solution for effective agricultural pest management.
Main Methods:
- A convolutional neural network (CNN) model, PestNet, was developed.
- PestNet integrates a multi-scale hybrid attention module with ResNet18.
- A comprehensive dataset of pest samples was compiled and used for training.
Main Results:
- PestNet achieved an average recognition accuracy of 93.61%.
- This represents a 2.04% improvement over the baseline ResNet18.
- Ablation experiments confirmed the positive impact of model modifications.
Conclusions:
- The developed PestNet model is effective for agricultural pest recognition.
- Deep learning shows significant potential for pest identification and management.
- The system offers an intelligent and convenient solution for farmers.
