PlantVillage Disease Detection Models
Multi-phase training pipeline for plant disease detection using the PlantVillage dataset.
Training Phases
- Phase 1: Multiclass classification (38 classes) - full dataset
- Phase 2: Balanced multiclass - class imbalance correction
- Phase 3: Binary classification (healthy vs diseased)
Architectures Tested
8 architectures of ViT, total to 24 models in three phases
Results
See individual phase directories for model weights, plots, and metrics.
