- Posed face recognition as a binary classification problem.
- Implemented one-shot learning for a face recognition problem.
- Applied the triplet loss function to learn a network's parameters in the context of face recognition.
- Mapped face images into 128-dimensional encodings using a pretrained model.
- Performed face verification and face recognition with these encodings.
This project requires the following Python libraries and packages:
tensorflow
numpy
pandas
PIL
keras
You can install these dependencies using pip:
pip install tensorflow numpy pandas pillow keras
- Clone the repository to your local machine:
git clone https://github.com/justinliu23/facial-recognition-system.git
- Navigate to the project directory:
cd face_recognition
- Ensure all dependencies are installed
- Launch Jupyter Notebook:
jupyter notebook
- Open the notebook
Face_Recognition.ipynb
. - Execute the cells in the notebook sequentially to run the face recognition pipeline.
Here is a brief guide on how to use the project:
In this section, you will compare two images pixel-by-pixel to determine if they are of the same person.
Using a Convolutional Neural Network (ConvNet), you will compute encodings for face images.
The Triplet Loss function helps learn parameters that improve the accuracy of the model in distinguishing between different faces.
You will load a pre-trained model to compute the 128-dimensional encodings for the face images.
- Face Verification: Determine if two images are of the same person (1:1 matching).
- Face Recognition: Identify the person in an image from a set of K persons (1:K matching).
- Input Shape: The input images are expected to have the shape (96, 96, 3).
- ConvNet Architecture: The model architecture includes convolutional layers, batch normalization, pooling layers, and dense layers.
- Loss Function: Triplet loss function is used to train the model.
- Data Format: The images should be in a format that can be processed by PIL (e.g., JPG, PNG).
- Dataset Structure: Organize your dataset such that each person has their own folder containing images of their face.