Idiap - Resnet V2 - MSCeleba

Inspired by **FaceNet** we here at Idiap trained our own CNN using the Inception Resnet 2 architecture using MSCeleba database. In this link you can find the script that trains this neural network.

To trigger this training it’s necessary to use the bob.learn.tensorflow package and run the following command:

$ ./bin/jman submit --name CELEB-GRAY --queue gpu -- bob_tf_train_generic MSCELEBA_inception_resnet_v2_center_loss_GRAY.py

Some quick details about this CNN (just as a mental note):

  • The hot encoded layer has 87662 neurons (number of identities in msceleba_182x_hand_prunned_44).

  • MSCeleba has a lot of mislabeling, a very simple prunning was implemented in this python package.

  • Faces were detected and croped to \(182 \times 182\) using MTCNN face and landmark detector

  • The following data augmentation strategies were implemented:

    • Random crop to \(160 \times 160\)

    • Random Flip

    • Images were normalized to have zero mean and standard deviation one

  • Learning rate of 0.1, 0.01, and 0.001

  • RMSProp as Optimizer

  • Batch size of 90

Two versions of it were trained: one providing color images for training and another one providing gray scale images.