metadata

title: AI Profile Picture Classification
emoji: 📚
colorFrom: red
colorTo: green
sdk: gradio
sdk_version: 4.22.0
app_file: app.py
pinned: false

AI vs. Real Profile Picture Classification

Overview

This project hosts a fine-tuned version of the Nahrawy/AIorNot model on Hugging Face's Model Hub, specifically adapted for distinguishing between AI-generated and real profile pictures. Utilizing transfer learning, the model has been fine-tuned on a diverse dataset comprising both real and synthetically generated face images to achieve high accuracy in classification tasks.

Model Description

The Nahrawy/AIorNot model, originally designed for image classification tasks, has been fine-tuned to specifically address the challenge of distinguishing AI-generated images from real ones. This adaptation makes it particularly suitable for applications in social media moderation, digital forensics, and any domain where such differentiation is critical.

Dataset

The model was fine-tuned on a combined dataset consisting of 20000 real and 8892 AI-generated profile pictures. The dataset was split into training and validation sets, ensuring a comprehensive learning and evaluation process.

Performance

The model's performance metrics, such as accuracy and F1 score, are reported as follows:

Accuracy: ...% F1 Score: ...%

Limitations and Bias

While the model demonstrates robust performance across diverse datasets, it's important to acknowledge potential limitations and biases inherent in the training data. Users are encouraged to evaluate the model in their specific contexts and consider these factors when interpreting the results.

Contributing

We welcome contributions to improve the model and its documentation. Whether you have suggestions for new features, bug reports, or requests, please feel free to open an issue or submit a pull request.

License

Apache 2.0

Acknowledgements

Special thanks to the creators of the Nahrawy/AIorNot model and the Hugging Face team for providing the infrastructure that makes this project possible. Additionally, this project is my Mid-Term assignment for the Computer Vision Class at Houston Community College (HCC).