Develop the Herbs Plant Classification Model for Patient Care Using Deep Learning

Master's Thesis - Bahir Dar University, Faculty of Computing · 2023

This research focuses on noise removal in the classification and description of herbal plants using the Xception model. The study aims to improve accuracy by addressing specific challenges such as shadow removal, dust removal, and color correction through the integration of advanced algorithms. The research incorporates the Shadow Detection and Removal (SDR) algorithm for shadow removal, the Top-hat transform algorithm for dust removal, and the Gray World Algorithm along with Contrast Limited Adaptive Histogram Equalization (CLAHE) for color correction and histogram enhancement. The evaluation demonstrates exceptional accuracy, achieving a classification accuracy of 99% and validation accuracy of 97.8%.

📊 Key Metrics

99%

Classification Accuracy

97.8%

Validation Accuracy

Xception

Model Architecture

3.43/4.0

Thesis Grade

🎯 Problem Statement

Traditional medicine relies heavily on accurate identification of herbal plants, but manual classification is time-consuming and error-prone. Healthcare applications need automated systems to classify and describe herbal plants accurately, supporting patient care and botanical research.

🔬 Methodology

We developed a deep learning classification system using the Xception architecture, integrated with advanced image preprocessing techniques. The pipeline includes Shadow Detection and Removal (SDR) algorithm, Top-hat transform for dust removal, and Gray World Algorithm with CLAHE for color correction and enhancement. The model was trained on a custom dataset of herbal plant images with extensive data augmentation.

💡 Key Contributions

→Comprehensive preprocessing pipeline addressing shadow, dust, and color issues
→Integration of multiple image enhancement algorithms for robust classification
→Achieved 99% classification accuracy and 97.8% validation accuracy
→Practical application for healthcare and traditional medicine support

🧪 Experiments & Evaluation

Collected and curated custom dataset of herbal plant images
Implemented and evaluated preprocessing algorithms (SDR, Top-hat, CLAHE)
Fine-tuned Xception model on preprocessed images
Conducted extensive data augmentation for generalization
Evaluated on separate validation set with cross-validation

📈 Results

The integrated system achieved exceptional performance with 99% classification accuracy and 97.8% validation accuracy. The preprocessing pipeline significantly improved model robustness by handling real-world image quality issues. The thesis received a grade of 3.43/4.0, demonstrating strong research contribution.

🚀 Future Work

Expand dataset to include more plant species
Deploy as mobile application for field identification
Integrate with traditional medicine knowledge base
Extend to other botanical classification tasks

📚 Citation (BibTeX)

@thesis{amsalu2023herbal,
  title={Develop the Herbs Plant Classification Model for Patient Care Using Deep Learning},
  author={Amsalu, Surafel},
  school={Bahir Dar University, Faculty of Computing},
  year={2023},
  type={Master's Thesis},
  advisor={Gebeyehu Belay (PhD)},
  grade={3.43/4.0}
}