How gagan-009/plant-disease-classification-transfer-learning works
This repository is currently a small project scaffold, not an implemented classification application. The README documents a planned PlantVillage transfer-learning study and pipeline; the tree contains split CSVs and a dataset-analysis notebook, but no training, evaluation, Grad-CAM, or Streamlit implementation. Those documented stages are shown as planned components with null paths. Edges reflect the README’s stated pipeline, not inspected runtime calls.
Main entry points: splits Train/val/test splits · 01_dataset_analysis.ipynb Dataset analysis
Components 13
Data
- PlantVillage dataset
- Data preparation
- Train/val/test splits
data/splits - Dataset analysis
notebooks/01_dataset_analysis.ipynb
Modeling
- Baseline CNN
- Feature extraction
- Fine-tuning
- ResNet18, ResNet50, EfficientNet-B0
Evaluation and demo
- Model evaluation
- Error analysis
- Grad-CAM
- Streamlit demonstration
Other components
- Demo user
Connections 15
- PlantVillage dataset → Data preparation: provides data
README.md - Data preparation → Train/val/test splits: creates splits
README.md - Train/val/test splits → Dataset analysis: analyzed in
- Train/val/test splits → Baseline CNN: feeds
README.md - Train/val/test splits → Feature extraction: feeds
README.md - Train/val/test splits → Fine-tuning: feeds
README.md - ResNet18, ResNet50, EfficientNet-B0 → Feature extraction: backbones for
README.md - ResNet18, ResNet50, EfficientNet-B0 → Fine-tuning: backbones for
README.md - Baseline CNN → Model evaluation: evaluated by
README.md - Feature extraction → Model evaluation: evaluated by
README.md - Fine-tuning → Model evaluation: evaluated by
README.md - Model evaluation → Error analysis: followed by
README.md - Error analysis → Grad-CAM: followed by
README.md - Grad-CAM → Streamlit demonstration: followed by
README.md - Demo user → Streamlit demonstration: uses