Plant Disease Classification & Deep Learning
Lightweight dual-stage feature refinement for black gram leaf disease classification using ConVITSE
This paper introduces ConViTSE, a lightweight hybrid deep learning architecture for black gram leaf disease classification, combining ConvMixer, Vision Transformer, and Squeeze and Excitation blocks for enhanced feature extraction and refinement. It aims to improve accuracy and computational efficiency.
Executive Impact
The ConViTSE model offers significant advancements in automated plant disease detection, providing high accuracy, computational efficiency, and strong generalization across diverse crops.
0
Peak Classification Accuracy on Black Gram
0
Lightweight Model Footprint
0
Rapid Inference Time Per Image
High
Generalization Across Rice, Maize, Wheat
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
ConViTSE: Hybrid Feature Extraction and Refinement
The ConViTSE model integrates ConvMixer, Vision Transformer (ViT), and Squeeze and Excitation (SE) blocks to leverage their complementary strengths. ConvMixer efficiently captures local spatial patterns and fine-grained textures, crucial for early disease signs. ViT then processes these features to understand global contextual relationships across the entire leaf. Finally, SE blocks, implemented as Local Channel Attention Refinement (LCAR) and Global Channel Attention Refinement (GCAR) modules, adaptively enhance feature representation at both local and global levels by emphasizing disease-relevant channels.
Unmatched Accuracy & Robust Generalization
ConViTSE achieves a leading classification accuracy of 99.30% on the black gram dataset. This high performance is maintained across different crop datasets, demonstrating robust cross-domain generalization with accuracies of 98.75% for rice, 98.20% for maize, and 95% for wheat. The dual-stage feature refinement ensures that the model effectively distinguishes visually similar disease symptoms and adapts to diverse agricultural environments.
Key Finding
99.30% Peak Classification Accuracy on Black GramEnterprise Process Flow
Blackgram Leaf Dataset & Preprocessing
→
ConvMixer for Local Feature Extraction (LCAR)
→
ViT for Global Feature Extraction (GCAR)
→
Global Average Pooling
→
Dense Classifier
→
Classification Results
Comparative Performance Analysis: ConViTSE vs. Traditional Models (Black Gram)
| Model | Accuracy (%) | Precision | Recall | F1-score |
|---|---|---|---|---|
| VGG19 | 95.32 | 0.95 | 0.93 | 0.92 |
| ResNet-50 | 96.11 | 0.96 | 0.94 | 0.95 |
| ConvMixer Only | 94.50 | 0.92 | 0.91 | 0.92 |
| ViT Only | 95.30 | 0.93 | 0.92 | 0.92 |
| ConViTSE (Ours) | 99.30 | 0.97 | 0.98 | 0.98 |
Real-time Disease Management in Agriculture
ConViTSE's lightweight design (0.756 million parameters, 2.89 MB model size, 0.04s inference time) makes it highly practical for real-time deployment on mobile devices and edge computing platforms. This enables farmers to monitor plant diseases instantly via their phones, supporting quick intervention strategies. Its ability to generalize across diverse crops (rice, corn, wheat) further enhances its utility for widespread adoption in precision agriculture, addressing a critical need for efficient disease management.
Calculate Your Potential ROI
Automated black gram disease detection can significantly reduce crop losses and manual inspection time, leading to higher yields and reduced labor costs for farmers. The model's high accuracy ensures reliable early detection, minimizing the spread of disease and optimizing resource allocation.
Estimated Annual Savings
$0
Hours Reclaimed Annually
0
Your AI Implementation Roadmap
Our proven process ensures a smooth transition to AI-powered plant disease detection, maximizing your investment and minimizing disruption.
Phase 1: Discovery & Strategy
Initial consultation to understand your specific agricultural needs, existing infrastructure, and disease challenges. We'll define clear objectives and outline a tailored AI strategy for your crops.
Phase 2: Customization & Deployment
ConViTSE model fine-tuning with your farm's specific data, integration with mobile devices or existing field monitoring systems, and rigorous testing in your environment.
Phase 3: Training & Ongoing Support
Comprehensive training for your team on using the ConViTSE system. Continuous monitoring, updates, and expert support to ensure optimal performance and adaptation to new disease threats.
Ready to Transform Your Agricultural Operations?
Automate disease detection, boost yields, and secure your harvests with our cutting-edge AI solutions. Let's discuss how ConViTSE can revolutionize your farm.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat