Face Swapper

This project is a Python-based application that allows you to perform facial swapping between two uploaded images. It uses OpenCV and dlib libraries for face detection, landmark extraction, and image transformation.

Features

Detects facial landmarks using dlib's pre-trained shape predictor model. Performs preprocessing and color correction to improve swapping quality. Supports swapping faces in both directions (Image1 → Image2 and Image2 → Image1). Displays the results interactively using matplotlib.

Requirements

Ensure you have the following libraries installed:

• opencv-python
• dlib
• numpy
• matplotlib

To install the necessary libraries, run:

pip install opencv-python dlib numpy matplotlib

Additionally, download the pre-trained shape predictor model for facial landmarks:

wget -nc http://dlib.net/files/shape_predictor_68_face_landmarks.dat.bz2 -O shape_predictor_68_face_landmarks.dat.bz2
bzip2 -dk shape_predictor_68_face_landmarks.dat.bz2

How to Use

1. Setup:
   • Save the script as face_swap.py.
   • Place the downloaded shape_predictor_68_face_landmarks.dat file in the same directory as the script. Execution: Launch the script using Python:

python face_swap.py

• When prompted, upload two images via the interface.

Output:

The application will process the uploaded images and display: Image1 → Image2 face-swapped result. Image2 → Image1 face-swapped result.

Functions Overview

Main Functions

• improved_face_swap(image1, image2):

• Swaps the face from image2 onto image1 with color correction and mask blending. imshow(title, image, size):

• Displays an image using matplotlib. -Helper Functions -get_landmarks(image):

• Detects facial landmarks using dlib. transformation_from_points(points1, points2):

• Calculates the affine transformation matrix to align the faces. get_face_mask(image, landmarks):

• Generates a mask for facial regions. warp_im(image, matrix, dshape):

• Applies a warp transformation to the input image. improved_color_correction(im1, im2, landmarks1):