## About the Dataset

The dataset contains **AI-generated anime-style portrait images** (front-facing headshots).  
Each image belongs to one of two classes:

- **0 → Masculine**
- **1 → Feminine**

The dataset was generated carefully using an anime diffusion model to keep:
✅ consistent face framing  
✅ clear facial features  
✅ clean backgrounds  
✅ diverse hair/eye colors, expressions, and lighting

---

## Dataset Structure

You will receive:

### ✅ Training data
- `train_images/` → training images  
- `train.csv` → image id and label

Format:
```javascript
id,label
feminine_00001.jpg,1
masculine_00002.jpg,0
```

### ✅ Validation data
- `val_images/`  
- `val.csv`

### ✅ Test data (for leaderboard submissions)
- `test_images/`
- `test.csv` → only ids

Format:
```javascript
test_00001.jpg
test_00002.jpg
```

---

## Goal

Build a classifier that can correctly predict anime gender from image portraits.

This competition was created because most existing models and datasets focus on **human gender detection**, and anime-style classification is a fun and different challenge.