santhalakshminarayana

files_in_dir=os.listdir()
curr_path=os.getcwd()

#get image file names in current directory
image_names=[]
conventions=['jpeg','png','jpg']
for file in files_in_dir:
	ext=file.split('.')[-1]
	if ext in conventions:
		image_names.insert(0,file)

santhalakshminarayana

#Find contours in image using (tree retrival method) for hierarchy
image_conts=[]
for img in thsh_images:
	contours,_=cv2.findContours(img.copy(),cv2.RETR_TREE,cv2.CHAIN_APPROX_SIMPLE)
	image_conts.append(contours)

#Look for maximum area contour which describes page/rectangle structure in image
max_area_conts=[]
for contour in image_conts:
	max_ind,max_area=None,0

santhalakshminarayana

#Draw closest four sided shape around maximum contour which is our 
#area of interest in image
approx_cont=[]
for ind in range(len(images_read)):
	epsilon=0.02*cv2.arcLength(image_conts[ind][max_area_conts[ind]],True)
	approx=cv2.approxPolyDP(image_conts[ind][max_area_conts[ind]],epsilon,True)
	approx_cont.append(np.squeeze(approx))

santhalakshminarayana

rect_images=[]
for ind in range(len(images_read)):
	#top-left,bottom-left,bottom-right,top-right
	tl,bl,br,tr=approx_cont[ind].tolist()
	top_width=np.sqrt((tl[0]-tr[0])**2 + (tl[1]-tr[1])**2)
	bottom_width=np.sqrt((bl[0]-br[0])**2 + (bl[1]-br[1])**2)
	left_height=np.sqrt((tl[0]-bl[0])**2 + (tl[1]-bl[1])**2)
	right_height=np.sqrt((tr[0]-br[0])**2 + (tr[1]-br[1])**2)
	width=int(max(top_width,bottom_width))
	height=int(max(left_height,right_height))

santhalakshminarayana

#Digitise image in black and white as a scanned document
digitised_image_names=[]
for ind in range(len(rect_images)):
	img_gray=cv2.cvtColor(rect_images[ind].copy(),cv2.COLOR_BGR2GRAY)
	th=threshold_local(img_gray.copy(),101,offset=10,method="gaussian")
	img_gray=(img_gray>th)
	imgg=Image.fromarray(img_gray)
	size=(images_read[ind].shape[0],images_read[ind].shape[1])
	imgg.resize(size)
	name=curr_path+"/digitised_"+image_names[ind].split('.')[0]+'.jpg'

santhalakshminarayana

#Convert all digitised images to pdf format
digitised_images=[]
for name in digitised_image_names:
	imgg=Image.open(name)
	digitised_images.append(imgg)
	name=curr_path+"/digitised_images"+'.pdf'
	if len(digitised_images)>1:
		digitised_images[0].save(name,save_all=True,append_images=digitised_images[1:],resolution=100.0)
	else:
		digitised_images[0].save(name)

santhalakshminarayana

import cv2
import dlib

# Load cnn_face_detector with 'mmod_face_detector'
dnnFaceDetector=dlib.cnn_face_detection_model_v1("mmod_human_face_detector.dat")

# Load image 
img=cv2.imread('path_to_image')

# Convert to gray scale

santhalakshminarayana

# Tensorflow version == 2.0.0
import numpy as np
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras.models import Sequential,Model
from tensorflow.keras.layers import ZeroPadding2D,Convolution2D,MaxPooling2D
from tensorflow.keras.layers import Dense,Dropout,Softmax,Flatten,Activation,BatchNormalization
from tensorflow.keras.preprocessing.image import load_img,img_to_array
from tensorflow.keras.applications.imagenet_utils import preprocess_input
import tensorflow.keras.backend as K

santhalakshminarayana

# Prepare Train Data
x_train=[]
y_train=[]
person_rep=dict()
person_folders=os.listdir(path+'/Images_crop/')
for i,person in enumerate(person_folders):
  person_rep[i]=person
  image_names=os.listdir('Images_crop/'+person+'/')
  for image_name in image_names:
    img=load_img(path+'/Images_crop/'+person+'/'+image_name,target_size=(224,224))

santhalakshminarayana

# Softmax regressor to classify images based on encoding 
classifier_model=Sequential()	
classifier_model.add(Dense(units=100,input_dim=x_train.shape[1],kernel_initializer='glorot_uniform'))		
classifier_model.add(BatchNormalization())		
classifier_model.add(Activation('tanh'))
classifier_model.add(Dropout(0.3))
classifier_model.add(Dense(units=10,kernel_initializer='glorot_uniform'))
classifier_model.add(BatchNormalization())
classifier_model.add(Activation('tanh'))
classifier_model.add(Dropout(0.2))