🎅 Detect Santa Claus!

On Christmas Eve, Santa Claus's workshop is busier than ever. Elves are preparing gifts, reindeer are polishing their bells, and Santa himself… appears in thousands of photos sent from all around the world.

The problem is that not all these images actually capture the real Santa Claus. Some show costumes, others figurines, advertisements, or people dressed up. To avoid delivering the wrong gifts, Santa needs your help!

Your task is to build a binary image classification model that decides whether an image contains Santa Claus or not.

🎄 Dataset

The dataset is provided as an archive containing images organized into two directories, along with the CSV files train.csv and test.csv.

There are two classes:

• Santa – the image contains Santa Claus (label = 1)
• Not Santa – the image does not contain Santa Claus (label = 0)

📁 Dataset Structure

dataset/
├── train/
│ ├── img_000001.jpg
│ ├── img_000002.jpg
│ └── ...
├── test/
│ ├── img_001234.jpg
│ ├── img_001235.jpg
│ └── ...
├── train.csv
└── test.csv

📄 CSV File Structure

train.csv

Contains the images used for training the model.

Columns:

• image_path – relative path to the image (e.g. train/img_000123.jpg)
• label
  • 1 → Santa Claus
  • 0 → Not Santa Claus

test.csv

Contains the images for which predictions must be generated.

Columns:

• image_path – relative path to the image

⚠️ Note: This file does NOT contain the label column. It must be predicted by your model.

🎯 Task Objective

For each image in the test set, your model must decide:

• 1 → the image contains Santa Claus
• 0 → the image does not contain Santa Claus

The results will be compared against the ground truth labels, which are kept secret by the organizers.

⚙️ Evaluation Criteria

The performance of your solution will be evaluated using the F1-score, a metric that combines precision and recall, suitable for binary classification problems.

Formula:

📨 Submission File Format

You must submit a CSV file with exactly two columns, in the following format:

🔹 Submission Example

image_path,label
test/img_001234.jpg,1
test/img_001235.jpg,0
test/img_001236.jpg,1

📝 Scoring System

• F1-score ≥ 0.95 → maximum score 🎁
• F1-score < 0.70 → 0 points
• Intermediate values are scored proportionally

❄️ Notes

• File names do not contain class information
• Image order is randomized
• Any machine learning method is allowed
• The goal is to teach the model to recognize visual patterns

🎅 Santa Claus is counting on you!
Help him identify the correct images and deliver gifts on time!