Related Experiment Video
Updated: Nov 17, 2025
13:19
Deep Neural Networks for Image-Based Dietary Assessment
Published on: March 13, 2021
9.7K
Improved Vision-Based Detection of Strawberry Diseases Using a Deep Neural Network
Byoungjun Kim1, You-Kyoung Han2, Jong-Han Park2
1Division of Computer Science and Engineering, Jeonbuk National University, Jeonju, South Korea.
Frontiers in Plant Science
|February 15, 2021
Summary
Early detection of strawberry diseases is crucial for crop quality. This study introduces an improved deep neural network (DNN) using PlantNet for automated robot-assisted disease detection, enhancing accuracy and addressing data limitations.
Area of Science:
- Agricultural Science
- Computer Vision
- Plant Pathology
Background:
- Early detection of plant diseases is vital for maintaining crop quality and farm productivity.
- Automated systems are increasingly important for efficient agricultural practices.
Purpose of the Study:
- To propose an improved vision-based method for early strawberry disease detection using a deep neural network (DNN).
- To integrate this DNN into an automated robot system for practical application.
Main Methods:
- Utilized a two-stage cascade disease detection model incorporating PlantNet, a DNN backbone pre-trained on the PlantCLEF dataset.
- Compared PlantNet's performance against a backbone pre-trained on a general ImageNet dataset.
- Implemented a human-like cascade detection strategy to enhance accuracy.
Main Results:
- PlantNet demonstrated superior performance, outperforming an ImageNet-pre-trained backbone by at least 3.2% in mean Average Precision (mAP).
- The cascade detector further improved accuracy by up to 5.25% mAP.
- The approach effectively addressed the challenge of limited annotated data by leveraging plant domain knowledge.
Conclusions:
- PlantNet provides a viable solution to the lack-of-annotated-data problem in plant disease detection through applied domain knowledge.
- The human-like cascade detection strategy significantly enhances the accuracy of automated strawberry disease detection systems.
