Upload README.md

README.md

@@ -43,14 +43,14 @@ The repository includes two parallel directory trees, each containing the same 1
 
 ```
 EuroCoinDataset/
-├── Dataset_Status/          # Organised by issuing country (23 sub-folders)
+├── country_dataset/          # Organised by issuing country (23 sub-folders)
 │   ├── Andorra/
 │   ├── Austria/
 │   ├── Belgium/
 │   ├── ...
 │   └── Vatican-City/
 │
-└── Dataset_Value/           # Organised by denomination (8 sub-folders)
+└── denomination_dataset/     # Organised by denomination (8 sub-folders)
     ├── 1-cent/
     ├── 2-cent/
     ├── 5-cent/
@@ -63,8 +63,8 @@ EuroCoinDataset/
 
 ### File naming convention
 
-- `Dataset_Status`: `<Country>_<index>.<ext>` (e.g., `Germany_42.jpg`)
-- `Dataset_Value`: `<denomination>-<Country>_<index>.<ext>` (e.g., `2-euro-Germany_42.jpg`)
+- `country_dataset`: `<Country>_<index>.<ext>` (e.g., `Germany_42.jpg`)
+- `denomination_dataset`: `<denomination>-<Country>_<index>.<ext>` (e.g., `2-euro-Germany_42.jpg`)
 
 Images are provided in `.jpg` and `.jpeg` format at their original resolution as retrieved from the web.
 
@@ -134,7 +134,7 @@ The dataset was used in the following peer-reviewed study to benchmark deep lear
 - Several CNN architectures were evaluated (custom CNNs and pre-trained models via transfer learning).
 - Best models achieved high top-1 accuracy on both country and denomination classification tasks.
 - The dataset was split into training, validation, and test sets with stratified sampling.
-- The dual-folder structure (`Dataset_Status` / `Dataset_Value`) allows straightforward use for either single-label or multi-label classification scenarios.
+- The dual-folder structure (`country_dataset` / `denomination_dataset`) allows straightforward use for either single-label or multi-label classification scenarios.
 
 ---