facial_recognition_only_code.py

# coding: utf-8

# In[16]:

from imutils import face_utils
import numpy as np
import imutils
import argparse
import dlib
import cv2
import random
from PIL import Image
# import the necessary packages
from imutils.face_utils import FaceAligner
from imutils.face_utils import rect_to_bb
import argparse
import imutils
import dlib
import cv2

#import matplotlib.pyplot as plt

# construct the argument parser and parse the arguments
#ap = argparse.ArgumentParser()
#ap.add_argument("-p", "--shape-predictor", required=True,help="path to facial landmark predictor")
#%tb ap.add_argument("-i", "--image", required=True,help="path to input image")
#%tb args = vars(ap.parse_args())



# In[17]:

detector = dlib.get_frontal_face_detector()
predictor = dlib.shape_predictor("shape_predictor_68_face_landmarks.dat")
fa = FaceAligner(predictor, desiredFaceWidth=256)

# Face Align Function
def faceAlign(image):
    #image = cv2.imread(img)
    image = imutils.resize(image, width=800)
    gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

    rects = detector(gray, 2)
    faceAligned = []
    for rect in rects:
        faceAligned = fa.align(image, gray, rect)
        #print(rect)
        #plt.imshow(faceAligned)
    
    return faceAligned


image = cv2.imread('dataset_cropped/Abhishek Bachan/AbhishekBachan71.jpg')
#plt.imshow(image)
#plt.imshow(faceAlign(image))


# In[18]:

import os                                                                                                             

def list_files(dir):
    r = []
    for root, dirs, files in os.walk(dir):
        for name in files:
            r.append(os.path.join(root, name))
    return r  

image_paths=list_files('dataset_cropped')
#print(image_paths)


# In[19]:

# Finding unique labesl and encoding them

all_labels = [imag.split('/')[-2] for imag in image_paths]

from sklearn import preprocessing
le = preprocessing.LabelEncoder()
le.fit(all_labels)
labels_enc = le.transform(all_labels)


# In[20]:

unique_labels = list(le.classes_)
print(unique_labels)
print(le.transform(unique_labels))


# In[21]:

from tqdm import tqdm

def read_images(image_paths):
    images = [];
    labels = [];
    for img_path in image_paths:
        image = faceAlign(cv2.imread(img_path));
        if image is not None:
            images.append(image)
            labels.append(img_path.split('/')[1])
    return images,labels


# In[22]:

## Encoding labels


#list(le.classes_)

def data_gen(batch_size=30):
    while True:
        image_paths=list_files('dataset_cropped')
        sample_dt=random.sample(image_paths,int(batch_size))
        
        images,labels=read_images(sample_dt)
        #images,measurements=random_append_augment_images(images,measurements)
        #print('ya')
        images=np.asarray(images)
        labels=np.asarray(labels)
        #print('images shape',images.shape)
        #print('measurements shape',measurements.shape)
        ind = np.random.choice(images.shape[0], int(batch_size), replace=False)
        images=images[ind,:,:,:]
        labels=le.transform(labels[ind])
        X=images
        y=labels
        
        yield X,y


# In[23]:

sample_length = 15
image_paths=list_files('dataset_cropped')
sample_dt=random.sample(image_paths,int(sample_length))
images,labels=read_images(sample_dt)


# In[14]:

# Alexnet

from keras.models import Sequential
from keras.layers import Flatten,Dense, Lambda
# There was a problem with original Keras progress bar due to which Notebook used to hang.
# Changing the progress bar with another version of it so as to get the code working in Notebook. Look at model.fit command on how its used
#from keras_tqdm import TQDMNotebookCallback 
from keras.layers import Conv2D, MaxPooling2D, Dropout, Activation, Cropping2D, ELU
from keras import backend as K
from keras.preprocessing.image import ImageDataGenerator
from keras.callbacks import ModelCheckpoint

input_shape=(image.shape[0],image.shape[1],3)
model=Sequential()
model.add(Lambda(lambda x:x/255.0-0.50,input_shape=input_shape))
model.add(Conv2D(3,1,1,subsample=(1,1),border_mode="valid",init='he_normal'))
model.add(ELU())
model.add(Conv2D(24,5,5,subsample=(2,2),border_mode="valid",init='he_normal'))
model.add(ELU())
model.add(Conv2D(36,5,5,subsample=(2,2),border_mode="valid",init='he_normal'))
model.add(ELU())
model.add(Dropout(.4))
model.add(Conv2D(48,5,5,subsample=(2,2),border_mode="valid",init='he_normal'))
model.add(ELU())
model.add(Conv2D(64,3,3,subsample=(1,1),border_mode="valid",init='he_normal'))
model.add(ELU())
model.add(Conv2D(64,3,3,subsample=(1,1),border_mode="valid",init='he_normal'))
model.add(Dropout(.3))
model.add(ELU())
model.add(Flatten())
model.add(Dense(1164,init='he_normal'))
model.add(Activation('relu'))
model.add(Dropout(.2))
model.add(Dense(100,init='he_normal'))
model.add(ELU())
model.add(Dense(50,init='he_normal'))
model.add(ELU())
model.add(Dense(10,init='he_normal'))
model.add(ELU())
model.add(Dense(1,init='he_normal'))

model.compile(loss='mse',optimizer='adam',metrics=['accuracy'])

batch_size=30
epochs=4
gen_train=data_gen(15)
gen_valid=data_gen(15)


# In[15]:

model.fit_generator(gen_train, samples_per_epoch=25,nb_epoch=epochs,validation_data=gen_valid,nb_val_samples=25, max_q_size=25, nb_worker=4, pickle_safe=True, verbose =2)


# In[ ]: