Face-Similarity-and-Attribute-Analysis-Using-DeepFace

This repository contains a Python script that uses the DeepFace library to perform face detection, embedding extraction, similarity comparison, and facial attribute analysis (emotion, age, gender, and race) between two images.

Features

• Face Detection: Detects faces in the provided images using DeepFace.
• Embedding Extraction: Extracts facial embeddings using a pre-trained deep learning model (e.g., Facenet) to represent each detected face numerically.
• Cosine Similarity Calculation: Computes the cosine similarity between two facial embeddings to determine how similar the faces are.
• DeepFace Verification: Leverages DeepFace's built-in verification to compute the distance between faces and converts it to a similarity percentage.
• Facial Analysis: Analyzes the detected face for attributes like emotion, age, gender, and race.
• Visualization: Displays the detected faces side by side for visual comparison.

How the Process Works

1. Face Detection:

   • The script starts by detecting faces in the two input images using DeepFace's extract_faces function. If a face is detected, it is returned for further analysis; otherwise, an appropriate message is displayed.

2. Embedding Extraction:

   • For each detected face, the script extracts facial embeddings using a deep learning model (e.g., Facenet) specified by the user. These embeddings are high-dimensional vectors that represent the facial features in a numerical format.

3. Similarity Calculation:

   • The script calculates the cosine similarity between the extracted embeddings of the two faces. Cosine similarity measures the angle between two vectors in a multi-dimensional space, which helps in understanding how similar or different the two faces are.
   • Two similarity scores are computed:
     • Custom Cosine Distance: A measure of distance between embeddings, converted to a linear similarity percentage.
     • DeepFace Distance: The default distance measure used by DeepFace, which is also converted to a linear similarity percentage.

4. Facial Analysis:

   • The script uses DeepFace's analyze function to determine various facial attributes like emotion, age, gender, and race for each detected face.

5. Visualization:

   • The detected faces are displayed side by side using Matplotlib, along with the analysis results printed in the console.

Requirements

To use this script, you need to install the following Python libraries:

• deepface
• matplotlib
• numpy
• scikit-learn

You can install all the required packages by running:

pip install deepface matplotlib numpy scikit-learn