twmht

Deep Learning for Face Recognition (May 2016)

Popular architectures

• FaceNet (Google)
  • They use a triplet loss with the goal of keeping the L2 intra-class distances low and inter-class distances high
• DeepID (Hong Kong University)
  • They use verification and identification signals to train the network. Afer each convolutional layer there is an identity layer connected to the supervisory signals in order to train each layer closely (on top of normal backprop)
• DeepFace (Facebook)
  • Convs followed by locally connected, followed by fully connected

twmht

 /usr/local/cuda/bin/nvcc -c file1.cu
  g++ -c file2.cpp
  g++ -o test file1.o file2.o -L/usr/local/cuda/lib64 -lcudart
  ./test

twmht

import numpy as np, h5py
from collections import defaultdict
import os

f = h5py.File('wider_face_train.mat','r')
event_list = f.get('event_list')
print f[event_list.value[0][0]].value.tostring()

#  for i in range(0, 61):
    #  print f[event_list.value[0][i]].value.tostring()

twmht

# -*- coding: utf-8 -*-
"""
Created on Wed Jul  6 10:13:27 2016

@author: gene
"""

import xml.etree.cElementTree as ET
import Image
import os, glob

twmht

import PIL
import cv2
from PIL import ImageFont
from PIL import Image
from PIL import ImageDraw
import string
import os
import shutil
import random
from collections import defaultdict

twmht

import glob
from collections import defaultdict
import cv2
import numpy as np
import random

char2num = {
    '0': 0, '1': 1, '2': 2, '3': 3, '4': 4, '5': 5, '6': 6, '7': 7, '8': 8,
    '9': 9, 'A': 10, 'B': 12, 'C': 13, 'D': 14, 'E': 15, 'F': 16, 'G': 17,
    'H': 18, 'I': 19, 'J': 20, 'K': 21, 'L': 23, 'M': 24, 'N': 25, 'O': 26,

twmht

from mnist import MNIST
import numpy as np
import lmdb
import caffe
mndata = MNIST('./data')

count = 0
env = lmdb.open('mnist_lmdb', map_size=1000*1000*1000)

txn = env.begin(write=True)

twmht

import os
import struct
from array import array
import numpy as np


class MNIST(object):
    def __init__(self, path='.'):
        self.path = path

twmht

from mnist import MNIST
import numpy as np
import cv2
import lmdb
import caffe
mndata = MNIST('./data')
images, labels = mndata.load_training()
labels = np.array(labels)
images = np.array(images).reshape(len(labels), 28, 28).astype(np.uint8)

twmht

import caffe
import numpy as np
# http://stackoverflow.com/questions/33828582/vgg-face-descriptor-in-python-with-caffe
img = caffe.io.load_image( "ak.png" )
img = img[:,:,::-1]*255.0 # convert RGB->BGR
avg = np.array([129.1863,104.7624,93.5940])
img = img - avg # subtract mean (numpy takes care of dimensions :)
img = img.transpose((2,0,1))
img = img[None,:] # add singleton dimension
net = caffe.Net("VGG_FACE_deploy.prototxt","VGG_FACE.caffemodel",  caffe.TEST)