abhayvashokan / live-face-recognition

Live training and recognition of faces using convolutional neural network.

Live-Face-Recognition

Live training and recognition of faces using convolutional neural network.

tkinter library is used for creating the user interface for the project.

During image generation, the head region of the new user is cropped using Haarcascade classifier. A Gaussian blur filter is applied and it is resized to 64x64. Finally the image is normalized bu dividing it with 255.

A custom CNN model is created with 2 convolution, 2 max-pooling and 3 dense layers due to limitations from the computational point of view for the training process. This model is used for face recognition during test.

Setup

pip install -r requirements.txt