-
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
-
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 *