Plant Disease Detection Using
Deep Learning
Team 7: Guide: Deepak D
Rahul K Prabhu
Sanchitha
Sarthak Pramod Pai
Sharan Raghuveer Pai
Problem Statement
In India, approximately 70% of the population relies on agriculture. Identifying plant diseases is crucial to prevent significant
losses in crop yields. However, manually observing plant diseases is a challenging task that demands a considerable amount
of labor, expertise in plant pathology, and an excessive amount of time. There is a pressing need for an automated system
that can accurately detect plant diseases using advanced technologies. Such a system would reduce the reliance on expert
inspections and chemical processes, making disease detection more affordable and scalable for large farms.
Existing Systems
• Manual Inspection: Labor-intensive and prone to human error.
• Chemical Analysis: Costly, environmentally invasive, and impractical for large-scale farms.
• Inconsistency and Accuracy: Manual inspections can lead to inconsistent and inaccurate
results.
• Economic Losses: Significant threat to agricultural productivity, leading to substantial
economic losses and reduced food security
Goals:
• Automated Disease Detection
• Enhanced Accuracy
• Cost-Effectiveness
• Real-Time Analysis
• User-Friendly Interface
Design Methadology
• Data Collection
• Preprocessing
• Feature Extraction
• Model Development
• System Integration
• Testing and Evaluation
Model and its Implementation
Model: Convolutional Neural Network (CNN)
CNN Architecture:
• Input Layer: Takes in the preprocessed image data.
• Convolutional Layers: Apply filters to extract feature maps.
• Pooling Layers: Reduce the dimensionality of the feature maps.
• Fully Connected Layers: Combine features to classify the images.
• Output Layer: Provides the probability of each disease class.
Model Training:
• Train the CNN using labeled training data.
• Use a large dataset to improve model generalization and accuracy.
Model Validation and Testing:
• Validate the model using a separate validation dataset to tune hyperparameters.
• Test the model on a test dataset to evaluate performance metrics like accuracy, precision, recall, and F1-score.
Supported Plants and Diseases
Apple: Corn :
• Apple scab • Cercospora leaf spot Gray leaf spot Tomato:
• Black rot • Common rust • Bacterial spot
• Cedar apple rust • Northern Leaf Blight • Early blight
• Healthy • Healthy • Late blight
• Leaf Mold
• Septoria leaf spot
Blueberry: Orange: • Spider mites Two-
• Healthy • Haunglongbing spotted spider mite
• Target Spot
• Yellow Leaf Curl Virus
• Mosaic virus
Cherry: Grape: • Healthy
• Powdery mildew • Black rot
• Healthy • Esca (Black Measles)
• Leaf blight (Isariopsis Leaf Spot) Strawberry
• Healthy • Leaf scorch
• Healthy
Supported Plants and Diseases
Peach: Raspberry:
• Bacterial spot • Healthy
• Healthy
Pepper (bell): Soybean:
• Bacterial spot • Healthy
• Healthy
Potato: Squash:
• Early blight • Powdery mildew
• Late blight
• Healthy
Result
Result
Result
Result
Result
Conclusion
Summary:
• Developed an automated plant disease detection system using image processing and machine learning.
• Addressed the limitations of manual inspection and chemical analysis by providing a cost-effective, accurate, and scalable solution.
Achievements:
• Automated Detection: Enabled real-time, automated detection of plant diseases.
• Enhanced Accuracy: Achieved high accuracy with advanced CNN models.
• User-Friendly: Created an accessible platform for farmers to monitor plant health easily.
Future Work:
• Expand the dataset to include more plant species and disease types.
• Improve the system’s robustness to varying environmental conditions.
• Integrate additional features such as disease treatment recommendations.
References
• R. G. de Luna, E. P. Dadios, and A. A. Bandala
• P. Revathi and M. Hemalatha
• J. J. Davis and D. C. Gilligan
• H. Durmus, E. O. Gunes, and M. Kirci
• S. Zhang, W. Huang, H. Zhang, H. Zhao, and Y. Wang
• M. Brahimi, K. Boukhalfa, and A. Moussaoui
