Face Recognition GUI

A desktop application for face recognition on photos and real-time webcam using Dlib and OpenCV, with an intuitive graphical interface for data collection, model training, and recognition.

Features

• Dual Recognition Modes:
  • Photo Recognition: Upload and recognize faces in images
  • Real-time Webcam: Live face recognition from webcam feed
• Integrated Data Collection: Collect training data through the GUI
• Model Training: Train and save recognition models with one click
• Haar Cascade Detection: Pre-trained frontal face detector
• GUI Workflow: Step-by-step guided interface for the entire pipeline
• Visual Results: Bounding boxes and labels on detected/recognized faces

Architecture

+------------------------------------------+
|         GUI Layer (Tkinter)              |
|  +-----------+ +---------------+          |
|  | Get Data  | | Train Model   |          |
|  | Recognize | | Webcam Feed   |          |
|  +-----------+ +---------------+          |
+------------------------------------------+
|       Face Detection (OpenCV)            |
|  - Haar Cascade (haarcascade_frontal     |
|    face_default.xml)                     |
|  - Multi-scale detection                 |
+------------------------------------------+
|    Face Recognition (Dlib + ML)          |
|  - Facial landmark detection             |
|  - Feature encoding (128-dim vector)     |
|  - Distance-based matching               |
+------------------------------------------+
|       Storage                            |
|  - dataset/ (collected face images)      |
|  - Models/ (trained model files)         |
+------------------------------------------+

Tech Stack

Category	Technology
Language	Python 3.8+
Face Detection	OpenCV (Haar Cascade)
Face Recognition	Dlib
ML	scikit-learn
GUI	Tkinter
Image Processing	OpenCV, NumPy
Notebook	Jupyter

Face Recognition Pipeline

Step 1: Face Detection

Uses OpenCV's pre-trained Haar Cascade classifier to detect frontal faces:

Input Image -> Grayscale -> Haar Cascade Detection -> Bounding Boxes -> Crop Faces

Parameter	Value	Description
scaleFactor	1.1	Image scale reduction at each level
minNeighbors	5	Detections needed to retain a rectangle
minSize	(30, 30)	Minimum possible face size

Step 2: Feature Extraction

Dlib extracts a 128-dimensional embedding vector for each detected face:

• Face is aligned using 68-point facial landmarks
• Normalized to a standard size and orientation
• Passed through a deep neural network to produce a 128-D encoding
• Faces of the same person produce similar encodings (small Euclidean distance)

Step 3: Recognition

Matching is performed by computing the Euclidean distance between face encodings:

distance = || encoding_1 - encoding_2 ||_2

Distance	Result	Interpretation
< 0.4	Match	Same person (high confidence)
0.4 - 0.6	Possible Match	Same person (medium confidence)
> 0.6	No Match	Different person

Step 4: Data Collection Workflow

Start -> Click "Get Data" -> Webcam activates -> Faces detected ->
Images saved to dataset/ -> Enough samples collected -> Click "Train Model" ->
Model saved to Models/ -> Ready for recognition

Pre-trained Models

Model	File	Purpose
Haar Cascade	haarcascade_frontalface_default.xml	Frontal face detection
Dlib Face Recognition	Models/	128-D face encoding model

Getting Started

Prerequisites

• Python 3.8 or higher
• Webcam (for real-time recognition)

Installation

git clone https://github.com/nntrivi2001/Face-recognition-GUI.git
cd Face-recognition-GUI
pip install opencv-python numpy dlib scikit-learn imutils

Usage

1. Run the application:

   jupyter notebook "Face Recognition/FaceRecognition.ipynb"

2. Collect Training Data:

   • Click the "Get Data" button
   • Stand in front of the webcam
   • Multiple face images will be captured and saved to dataset/

3. Train the Model:

   • Click "Train Model" or "Train and recognize"
   • The model processes all images in dataset/
   • Trained model is saved to Models/

4. Recognize Faces:

   • Photo Mode: Choose a picture for recognition
   • Webcam Mode: Enable real-time webcam recognition

Project Structure

Face-recognition-GUI/
|-- Face Recognition/
|   |-- FaceRecognition.ipynb   # Main application with GUI
|-- Models/                     # Trained model files
|-- haarcascade_frontalface_default.xml   # Pre-trained Haar Cascade
|-- dataset/                    # Collected face images
|-- image.png                   # Screenshot 1
|-- image-1.png                 # Screenshot 2
|-- .gitignore
|-- .gitattributes
|-- README.md

Screenshots