Poject Description

Build a face recognition system

Goals

• Implement the triplet loss function
• Use a pretrained model to map face images into 128-dimensional encodings
• Use these encodings to perform face verification and face recognition

Packages used

• keras.models - Sequential-
• keras.layers - Conv2D, ZeroPadding2D, Activation, Input, concatenate -
• keras.models - Model -
• keras.layers.normalization - BatchNormalization -
• keras.layers.pooling - MaxPooling2D, AveragePooling2D -
• keras.layers.merge - Concatenate -
• keras.layers.core - Lambda, Flatten, Dense -
• keras.initializers - glorot_uniform -
• keras.engine.topology - Layer -
• keras - backend - K.set_image_data_format('channels_first')
• cv2
• os
• numpy
• pandas
• tensorflow
• fr_utils import *
• inception_blocks_v2 import *

Used data set

details inside the notebook