A dataset of euro coin images classified by denomination
The data set is structured suitable to read using Keras ImageDataGenerator.flow_from_directory()
method.
.
├── extract.py # Python script used to extract coins from images
├── src # Source images
| └── ...
└── classified # Classified example images
├── train # Training set, structured according to denomination
| ├── 2c
| └── ...
└── test # Test or validation set
└── ...
I'd be happy to accept corrections or additions to the dataset. Please make sure to follow the general style and dimensions of the existing data.
Source images are shot top-down on an A4 sized white background with soft ambient lighting. The source images
are then post-processed using the supplied extract.py
file to isolate individual coins and normalize the sizes
before manual classification.
This project is licensed under the MIT License - see the LICENSE.md file for details