nude-classifier-2d-3d 1.0.2

A Python package for classifying nude images as real (3D) or drawn (2D) using a pre-trained EfficientNet-B0 model.

Nude Classifier

A Python package for binary classification of images to determine if they are real (3D) or drawn (2D) nude images using a pre-trained EfficientNet-B0 model.

Installation

Using pip

pip install nude-classifier

Using Poetry

poetry add nude-classifier

Requirements

• Python 3.8+
• PyTorch
• torchvision
• PIL (Pillow)

See requirements.txt or pyproject.toml for a full list of dependencies.

Usage

from nude_classifier_2d_3d import ClassifyNudeAs2dOr3d

# Initialize the classifier
classifier = ClassifyNudeAs2dOr3d()

# Classify an image
image_path = "path/to/your/image.jpg"
prob_3d = classifier.is_3d_nude_prob(image_path)
prob_2d = classifier.is_2d_nude_prob(image_path)

print(f"Probability of 3D (real) nude: {prob_3d:.2f}")
print(f"Probability of 2D (drawn) nude: {prob_2d:.2f}")

Model Details

• Architecture: EfficientNet-B0
• Input: RGB images (will be resized to 224x224 pixels)
• Output: Probability score for 3D (real) nude images
• Pre-trained Model: Included in the package (models/best_efficientnet_b0.pth)

License

This project is licensed under the MIT License. See the LICENSE file for details.

Disclaimer

This package is intended for research purposes only. Use responsibly and in compliance with applicable laws and regulations.