Related Experiment Video
Updated: May 30, 2025
08:47
Author Spotlight: UAV Remote Sensing for Efficient Invasive Plant Biomass Estimation
Published on: February 9, 2024
1.3K
Development of a handheld GPU-assisted DSC-TransNet model for the real-time classification of plant leaf disease
Midhun P Mathew1, Sudheep Elayidom1, V P Jagathy Raj2
1CS Division, SOE-Cochin University of Science and Technology, Cochin, Kerala, India.
Scientific Reports
|January 28, 2025
Summary
A new hybrid deep learning model, DSC-TransNet, accurately identifies plant leaf diseases in real-time. This advancement aids sustainable agriculture by enabling rapid detection and intervention, improving crop yields.
Area of Science:
- Agricultural Science
- Computer Science
- Deep Learning
Background:
- Accurate leaf disease identification is vital for sustainable agriculture and food security.
- Existing methods may lack the precision or real-time capabilities needed for effective crop management.
Purpose of the Study:
- To develop a hybrid deep learning model for accurate, real-time classification of plant leaf diseases.
- To enhance the detection of diseases in grape, bell pepper, and tomato plants.
Main Methods:
- A hybrid model combining VGG19 features with transformer encoder blocks was developed.
- Depthwise separable convolutional (DSC) layers were integrated for computational efficiency.
- The model, DSC-TransNet, was trained and validated on diverse datasets and tested on NVIDIA Jetson Nano.
Main Results:
- The DSC-TransNet model achieved high performance metrics, including 99.97% accuracy, 99.94% precision, recall, sensitivity, and F1-score, and 0.98 AUC for grape leaves.
- The model demonstrated effectiveness across datasets including bell pepper and tomato.
- Integration of DSC layers improved computational efficiency without sacrificing performance.
Conclusions:
- The DSC-TransNet model offers a reliable tool for automated plant disease classification.
- This research contributes to agricultural sustainability through enhanced crop monitoring and management.
- The model's real-time capabilities support timely interventions, mitigating economic losses and improving crop yields.
