README.md

---
{}
---

# ViPer: Metric
GitHub: https://github.com/sogandstorme/ViPer_Personalization

## Example

```bash
git clone https://github.com/sogandstorme/ViPer_Personalization.git
cd ViPer_Personalization
```

```python
from metric import (
    set_device,
    load_context_images,
    initialize_processor_and_model,
    calculate_score
)

# Ensure that the number of liked and disliked images are the same

negative_image_paths = [
    "disliked/0.png",
    "disliked/1.png",
    "disliked/2.png",
    "disliked/3.png",
    "disliked/4.png",
    "disliked/5.png",
    "disliked/6.png",
    "disliked/7.png",
    "disliked/8.png",
]

positive_image_paths = [
    "liked/0.png",
    "liked/1.png",
    "liked/2.png",
    "liked/3.png",
    "liked/4.png",
    "liked/5.png",
    "liked/6.png",
    "liked/7.png",
    "liked/8.png",
]

# Specify the address of the query image
query_image = "query.png"

device = set_device("cuda:0")
    
# Initialize processor and model
device = set_device("cuda:0")
context_images = load_context_images(negative_image_paths, positive_image_paths)
processor, model = initialize_processor_and_model(device)

# Calculate and print score
score = calculate_score(processor, model, context_images, query_image)

print(score)
```