Enhancing Deep Learning Model Performance through Data Generation Strategies: Classification of Olive Tree Diseases
Summary
- Researchers propose a comprehensive data generation strategy to address insufficient data and class imbalance in plant disease diagnosis.
- The approach combines traditional augmentation, CutMix, MixUp, and DCGANs for synthetic image generation.
- Generated data is evaluated using three scenarios: balancing classes, increasing sample counts, and expanding while maintaining balance.
- Experiments using the EfficientNetV2B0 architecture demonstrate substantial improvements in classification accuracy.
Why it matters
This research provides practical data generation and distribution strategies to improve deep learning classification performance in data-constrained domains like precision agriculture.