DSP_Project

This repository is dedicated to the data science practice project titled 'Understanding Real and Fake Faces'.

Dataset

The dataset consists of 1081 real images and three categories of a total of 960 fake images—easy, medium, and hard to detect.

Dataset: https://www.kaggle.com/datasets/ciplab/real-and-fake-face-detection

Directory Structure

├── Augmentation_ds.ipynb          # For apply face croping and augmentation on images
├── CNN.py                         # Proposed CNN model
├── CNN_Feature_Extraction.ipynb   # For featuremap extraction from each images
├── Classification.py              # For classification using lazypredict
├── EDA.py                         # For exploratory data analysis
├── Eigen_faces.ipynb              # For generating eigen faces
├── LIME_Explainer.py              # XAI using LIME
├── SHAP_Explainer.py              # XAI using SHAP

Methodology & Deliverables

1. Workflow

2. SHAP Explainer

3. LIME Explainer