Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/arsath-eng/face_classification

a deep learning project that uses ResNet and Inception architectures to classify real vs AI-generated face images. The project includes two models trained on a custom dataset, achieving validation accuracies of 52.45% (ResNet) and 52.94% (Inception). Built with TensorFlow and Flask, and a web interface for real time face classification
https://github.com/arsath-eng/face_classification

ai-detection computer-vision deep-learning face-detection flask image-classification machine-learning tensorflow

Last synced: 4 days ago
JSON representation

Host: GitHub
URL: https://github.com/arsath-eng/face_classification
Owner: arsath-eng
License: mit
Created: 2024-07-13T22:17:51.000Z (4 months ago)
Default Branch: main
Last Pushed: 2024-10-29T11:11:11.000Z (18 days ago)
Last Synced: 2024-10-29T13:12:11.170Z (18 days ago)
Topics: ai-detection, computer-vision, deep-learning, face-detection, flask, image-classification, machine-learning, tensorflow
Language: Jupyter Notebook
Homepage: https://huggingface.co/spaces/arsath-sm/face_classification
Size: 1.23 MB
Stars: 0
Watchers: 1
Forks: 0
Open Issues: 0
Metadata Files:
- Readme: README.md

Awesome Lists containing this project

README

        # Face Classification Project 🎭

[![Python](https://img.shields.io/badge/Python-3.9-blue.svg)](https://www.python.org)

[![Flask](https://img.shields.io/badge/Flask-2.1.0-green.svg)](https://flask.palletsprojects.com/)

[![TensorFlow](https://img.shields.io/badge/TensorFlow-latest-orange.svg)](https://tensorflow.org)

[![HuggingFace](https://img.shields.io/badge/🤗%20HuggingFace-Spaces-yellow.svg)](https://huggingface.co/spaces/arsath-sm/face_classification)

A deep learning project for detecting and classifying real and artificially generated face images using ResNet and Inception architectures. 👤✨



  



## 📋 Table of Contents

- [Overview](#overview)

- [Model Architecture](#model-architecture)

- [Performance](#performance)

- [Installation](#installation)

- [Usage](#usage)

- [Model Links](#model-links)

- [Future Improvements](#future-improvements)

## 🔍 Overview

This project implements two deep learning models to classify images as either real or AI-generated faces:

- **Model 1**: ResNet-based architecture with residual connections

- **Model 2**: Inception-style network with multi-scale feature processing

## 🏗️ Model Architecture

### ResNet Model (Model 1)

- Deep CNN with residual connections

- Handles vanishing gradient problem

- Effective for complex feature learning

- Multiple residual blocks with increasing filter sizes

- Dropout layers for regularization

### Inception Model (Model 2)

- Multi-scale feature processing

- Efficient computational design

- Parallel convolutional paths

- Adaptive to varying face sizes and orientations

- Better generalization capabilities

## 📊 Performance

| Model | Validation Accuracy | Validation Loss |

|-------|-------------------|----------------|

| ResNet | 52.45% | 0.7246 |

| Inception | 52.94% | 0.6913 |



  



## 🚀 Installation

### 1. Set up Conda Environment

```bash

# Create new environment

conda create -p face python=3.9

# Activate environment

conda activate face

# Install requirements

pip install -r requirements.txt

```

### 2. Requirements

Create a `requirements.txt` file with:

```

Flask==2.1.0

tensorflow

numpy

opencv-python

pillow

```

## 💻 Usage



  





  





  



### Running the Flask App

```bash

python app.py

```

Visit `http://localhost:5000` in your web browser to access the application.

## 🔗 Model Links

- [Model 1 on HuggingFace](https://huggingface.co/arsath-sm/face_classification_model1/tree/main)

- [Model 2 on HuggingFace](https://huggingface.co/arsath-sm/face_classification_model2/tree/main)

- [Demo Application](https://huggingface.co/spaces/arsath-sm/face_classification)

- [Flask Application Source](https://github.com/arsath-eng/face_classification.git)

## 🔄 Future Improvements

1. 📈 Data Quality

   - Investigate dataset biases

   - Enhance data diversity

2. 🔧 Model Architecture

   - Experiment with hybrid models

   - Implement transfer learning

   - Test ensemble methods

3. ⚡ Performance

   - Extended training periods

   - Hyperparameter optimization

   - Advanced data augmentation techniques

## 🔗 Dataset

- [dataset about real-fake human face classification](https://drive.google.com/file/d/16y2xEwwuf1v_W0BUDv1thzpNu_dnaJZ1/view?usp=sharing)

## 👥 Contributors

- Arsath S.M

- Faahiht K.R.M

- Arafath M.S.M

## 📄 License

[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)

This project is licensed under the [MIT License](LICENSE) - see the [LICENSE](LICENSE) file for details.

---



Made with ❤️ at University of Jaffna Faculty of Engineering