Better code quality, less dependency. (NVIDIA#12)

* Removed the dependency on opencv

The project's dependencies on opencv are sparse and this makes `Propagator`'s logic consistent with `smooth_filter`.

* Redundant code removed.

* Duplicated code in examples merged. Better timing.

USAGE.md

@@ -5,7 +5,7 @@ Licensed under the CC BY-NC-SA 4.0 license (https://creativecommons.org/licenses
 
 ### Setup
 
-We only test our code in the following environment. 
+We only tested our code in the following environment. 
 - OS: Ubuntu 16.04
 - CUDA: 9.1
 - **Python 2 from Anaconda2**

demo.py

@@ -2,20 +2,11 @@
 Copyright (C) 2018 NVIDIA Corporation.  All rights reserved.
 Licensed under the CC BY-NC-SA 4.0 license (https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode).
 """
-import os
-import torch
-from torch.autograd import Variable
-import torchvision.transforms as transforms
-import torchvision.utils as utils
+
 import argparse
-import time
-import numpy as np
-import cv2
-from PIL import Image
-from photo_wct import PhotoWCT
-from photo_smooth import Propagator
-from smooth_filter import smooth_filter
 
+import process_stylization
+from photo_wct import PhotoWCT
 
 parser = argparse.ArgumentParser(description='Photorealistic Image Stylization')
 parser.add_argument('--vgg1', default='./models/vgg_normalised_conv1_1_mask.t7', help='Path to the VGG conv1_1')
@@ -37,48 +28,13 @@
 
 # Load model
 p_wct = PhotoWCT(args)
-p_pro = Propagator()
 p_wct.cuda(0)
 
-content_image_path = args.content_image_path
-content_seg_path = args.content_seg_path
-style_image_path = args.style_image_path
-style_seg_path = args.style_seg_path
-output_image_path = args.output_image_path
-
-# Load image
-cont_img = Image.open(content_image_path).convert('RGB')
-styl_img = Image.open(style_image_path).convert('RGB')
-try:
-  cont_seg = Image.open(content_seg_path)
-  styl_seg = Image.open(style_seg_path)
-except:
-  cont_seg = []
-  styl_seg = []
-
-cont_img = transforms.ToTensor()(cont_img).unsqueeze(0)
-styl_img = transforms.ToTensor()(styl_img).unsqueeze(0)
-cont_img = Variable(cont_img.cuda(0), volatile=True)
-styl_img = Variable(styl_img.cuda(0), volatile=True)
-
-cont_seg = np.asarray(cont_seg)
-styl_seg = np.asarray(styl_seg)
-
-start_style_time = time.time()
-stylized_img = p_wct.transform(cont_img, styl_img, cont_seg, styl_seg)
-end_style_time = time.time()
-print('Elapsed time in stylization: %f' % (end_style_time - start_style_time))
-utils.save_image(stylized_img.data.cpu().float(), output_image_path, nrow=1)
-
-start_propagation_time = time.time()
-out_img = p_pro.process(output_image_path, content_image_path)
-end_propagation_time = time.time()
-print('Elapsed time in propagation: %f' % (end_propagation_time - start_propagation_time))
-cv2.imwrite(output_image_path, out_img)
-
-start_postprocessing_time = time.time()
-out_img = smooth_filter(output_image_path, content_image_path, f_radius=15, f_edge=1e-1)
-end_postprocessing_time = time.time()
-print('Elapsed time in post processing: %f' % (end_postprocessing_time - start_postprocessing_time))
-
-out_img.save(output_image_path)
+process_stylization.stylization(
+    p_wct=p_wct,
+    content_image_path=args.content_image_path,
+    style_image_path=args.style_image_path,
+    content_seg_path=args.content_seg_path,
+    style_seg_path=args.style_seg_path,
+    output_image_path=args.output_image_path,
+)