Pre-processing
- Auto-orientation: Pixel data oriented correctly, with EXIF-orientation stripped.
- Resize: Images resized to 640×640 pixels, stretched to fit.
Augmentation
Each image is augmented to create three versions, including:
- Horizontal Flip: 50% probability.
- Random Crop: 0-20% of the image.
- Random Shear: -10° to +10° horizontally and vertically.
- Brightness Adjustment: -15% to +15%.
- Exposure Adjustment: -10% to +10%.
- Salt and Pepper Noise: Applied to 0.1% of pixels.
Rock Types
- Basalt
- Chert
- Clay
- Conglomerate
- Diatomite
- Gypsum
- Olivine-basalt
- Shale (Mudstone)
- Siliceous Sinter
Usage
This dataset is ideal for training machine learning models in geosciences, education, and industrial applications, providing a comprehensive tool for accurate rock type classification.
Applications
- Geological Surveys: Assist in geological mapping and analysis.
- Educational Tools: Used in teaching materials for geology courses.
- Industrial Applications: Helps in the identification and processing of mineral resources.