## Purpose & Audience

This competition was created as part of the **Deep Learning for Computer Vision** course at **the University of Tartu**. 

## Task

Your goal is to train a computer vision model that accurately predicts which specific coin is depicted in each image. The coin variety is substantial: from euros of different national sides to Estonian kroons, Pakistani rupees, and beyond.

Most images feature **one coin**, though in some cases, both sides of that single coin are visible. This makes the classification more nuanced because details on either the obverse or reverse might be critical to disambiguate among similarly shaped coins.

---

## Dataset

We’ve provided the dataset split into **train** and **test** sets:

- **Train**: ~10,400 images (85% of the dataset).
- **Test**: ~1,288 images (15% of the dataset).

Each image is labeled with one of **315 classes**, combining **country + currency + denomination** into a single identifier. The dataset was assembled through:
- **Web scraping** across various coin-collector resources
- The coin image dataset [WanderDust’s Coin Images](https://www.kaggle.com/datasets/wanderdust/coin-images/data)

### Challenges & Caveats

- **Class Imbalances**: Some coins have many samples, while others are relatively underrepresented.
- **Mint Variations**: Coins of the same denomination might differ subtly due to different mints or yearly revisions.
- **Type Variations**: Coins of the same denomination might have different types.
- **Varied Conditions**: Tarnish, rust, and scratches can alter a coin’s appearance. Lighting and angle variations further increase complexity.
- **Subtle Differences**: Denomination of euro coins can only be distinguished by edge patterns.


## Getting Started

- **train.csv**: Lists each training image (by `Id`) and its `Class` label.  
- **test.csv**: Lists the test image IDs (no labels).  
- **sample_submission.csv**: Shows the required format for submissions (you’ll predict the coin class for each `Id`).  File names in this case are Ids. **Make sure not to add file extension to submission**

To participate, just:
1. Download the data.
2. Build your classification model.
3. Generate predictions (a CSV with each `Id` and your predicted `Class`).
4. Submit your CSV.