Updated the code to work on TensorFlow 2.0x and Python 3.

main.py

@@ -1,13 +1,13 @@
 from src import *
 
-gx, gy, gxlist, gylist = estimate_watermark('images/fotolia_processed')
+gx, gy, gxlist, gylist = estimate_watermark('Resized')
 
 # est = poisson_reconstruct(gx, gy, np.zeros(gx.shape)[:,:,0])
 cropped_gx, cropped_gy = crop_watermark(gx, gy)
 W_m = poisson_reconstruct(cropped_gx, cropped_gy)
 
 # random photo
-img = cv2.imread('images/fotolia_processed/fotolia_137840645.jpg')
+img = cv2.imread('Resized/image0000.jpg')
 im, start, end = watermark_detector(img, cropped_gx, cropped_gy)
 
 # plt.imshow(im)
@@ -16,7 +16,7 @@
 # W_m is the cropped watermark
 num_images = len(gxlist)
 
-J, img_paths = get_cropped_images('images/fotolia_processed', num_images, start, end, cropped_gx.shape)
+J, img_paths = get_cropped_images('Resized', num_images, start, end, cropped_gx.shape)
 # get a random subset of J
 idx = [389, 144, 147, 468, 423, 92, 3, 354, 196, 53, 470, 445, 314, 349, 105, 366, 56, 168, 351, 15, 465, 368, 90, 96, 202, 54, 295, 137, 17, 79, 214, 413, 454, 305, 187, 4, 458, 330, 290, 73, 220, 118, 125, 180, 247, 243, 257, 194, 117, 320, 104, 252, 87, 95, 228, 324, 271, 398, 334, 148, 425, 190, 78, 151, 34, 310, 122, 376, 102, 260]
 idx = idx[:25]
@@ -29,13 +29,13 @@
 C, est_Ik = estimate_blend_factor(J, Wm, alph)
 
 alpha = alph.copy()
-for i in xrange(3):
+for i in range(3):
 	alpha[:,:,i] = C[i]*alpha[:,:,i]
 
 Wm = Wm + alpha*est_Ik
 
 W = Wm.copy()
-for i in xrange(3):
+for i in range(3):
 	W[:,:,i]/=C[i]
 
 Jt = J[:25]

resize.py

@@ -0,0 +1,23 @@
+import os,sys
+from cv2 import cv2
+import glob
+from PIL import Image
+
+os.mkdir('Resized')
+path = "wildehm/"
+dirs = os.listdir( path )
+
+def resize():
+    i=0
+    for item in dirs:
+        if os.path.isfile(path+item):
+            image = cv2.imread(path+item)
+            imgResized = cv2.resize(image, (800,500))
+            cv2.imwrite("Resized/image%04i.jpg" %i, imgResized)
+            i +=1
+            cv2.imshow('image', imgResized)
+            cv2.waitKey(30)
+    cv2.destroyAllWindows()
+
+resize()
+

src/__init__.py

@@ -1,4 +1,4 @@
-from estimate_watermark import *
-from preprocess import *
-from image_crawler import *
-from watermark_reconstruct import *
+from .estimate_watermark import *
+from .preprocess import *
+from .image_crawler import *
+from .watermark_reconstruct import *

src/estimate_watermark.py

@@ -31,8 +31,8 @@ def estimate_watermark(foldername):
 
 	# Compute gradients
 	print("Computing gradients.")
-	gradx = map(lambda x: cv2.Sobel(x, cv2.CV_64F, 1, 0, ksize=KERNEL_SIZE), images)
-	grady = map(lambda x: cv2.Sobel(x, cv2.CV_64F, 0, 1, ksize=KERNEL_SIZE), images)
+	gradx = list(map(lambda x: cv2.Sobel(x, cv2.CV_64F, 1, 0, ksize=KERNEL_SIZE), images))
+	grady = list(map(lambda x: cv2.Sobel(x, cv2.CV_64F, 0, 1, ksize=KERNEL_SIZE), images))
 
 	# Compute median of grads
 	print("Computing median gradients.")
@@ -114,7 +114,7 @@ def poisson_reconstruct(gradx, grady, kernel_size=KERNEL_SIZE, num_iters=100, h=
 	est[1:-1, 1:-1, :] = np.random.random((m-2, n-2, p))
 	loss = []
 
-	for i in xrange(num_iters):
+	for i in range(num_iters):
 		old_est = est.copy()
 		est[1:-1, 1:-1, :] = 0.25*(est[0:-2, 1:-1, :] + est[1:-1, 0:-2, :] + est[2:, 1:-1, :] + est[1:-1, 2:, :] - h*h*laplacian[1:-1, 1:-1, :])
 		error = np.sum(np.square(est-old_est))
@@ -176,7 +176,7 @@ def watermark_detector(img, gx, gy, thresh_low=200, thresh_high=220, printval=Fa
 	if printval:
 		print(index)
 
-	x,y = (index[0]-rect[0]/2), (index[1]-rect[1]/2)
+	x,y = (index[0]-rect[0]//2), (index[1]-rect[1]//2)
 	im = img.copy()
 	cv2.rectangle(im, (y, x), (y+rect[1], x+rect[0]), (255, 0, 0))
 	return (im, (x, y), (rect[0], rect[1]))

src/watermark_reconstruct.py

@@ -4,8 +4,8 @@
 import scipy
 from scipy.sparse import *
 from scipy.sparse import linalg
-from estimate_watermark import *
-from closed_form_matting import *
+from .estimate_watermark import *
+from .closed_form_matting import *
 from numpy import nan, isnan
 
 def get_cropped_images(foldername, num_images, start, end, shape):
@@ -127,7 +127,7 @@ def estimate_normalized_alpha(J, W_m, num_images=30, threshold=170, invert=False
 
     print("Estimating normalized alpha using %d images."%(num_images))
     # for all images, calculate alpha
-    for idx in xrange(num_images):
+    for idx in range(num_images):
         imgcopy = thr
         alph = closed_form_matte(J[idx], imgcopy)
         alpha[idx] = alph
@@ -141,7 +141,7 @@ def estimate_blend_factor(J, W_m, alph, threshold=0.01*255):
     gx_jm = np.zeros(J.shape)
     gy_jm = np.zeros(J.shape)
 
-    for i in xrange(K):
+    for i in range(K):
         gx_jm[i] = cv2.Sobel(Jm[i], cv2.CV_64F, 1, 0, 3)
         gy_jm[i] = cv2.Sobel(Jm[i], cv2.CV_64F, 0, 1, 3)
 
@@ -153,7 +153,7 @@ def estimate_blend_factor(J, W_m, alph, threshold=0.01*255):
     estIk_grad = np.sqrt(gx_estIk**2 + gy_estIk**2)
 
     C = []
-    for i in xrange(3):
+    for i in range(3):
         c_i = np.sum(Jm_grad[:,:,:,i]*estIk_grad[:,:,i])/np.sum(np.square(estIk_grad[:,:,i]))/K
         print(c_i)
         C.append(c_i)
@@ -181,15 +181,15 @@ def solve_images(J, W_m, alpha, W_init, gamma=1, beta=1, lambda_w=0.005, lambda_
     sobely = get_ySobel_matrix(m, n, p)
     Ik = np.zeros(J.shape)
     Wk = np.zeros(J.shape)
-    for i in xrange(K):
+    for i in range(K):
         Ik[i] = J[i] - W_m
         Wk[i] = W_init.copy()
 
     # This is for median images
     W = W_init.copy()
 
     # Iterations
-    for _ in xrange(iters):
+    for _ in range(iters):
 
         print("------------------------------------")
         print("Iteration: %d"%(_))
@@ -208,7 +208,7 @@ def solve_images(J, W_m, alpha, W_init, gamma=1, beta=1, lambda_w=0.005, lambda_
         alpha_diag = diags(alpha.reshape(-1))
         alpha_bar_diag = diags((1-alpha).reshape(-1))
 
-        for i in xrange(K):
+        for i in range(K):
             # prep vars
             Wkx = cv2.Sobel(Wk[i], cv2.CV_64F, 1, 0, 3)
             Wky = cv2.Sobel(Wk[i], cv2.CV_64F, 0, 1, 3)
@@ -302,4 +302,4 @@ def changeContrastImage(J, I):
     cI2 = I[-1,-1, :]
 
     I_m = cJ1 + (I-cI1)/(cI2-cI1)*(cJ2-cJ1)
-    return I_m
+    return I_m