add hd-celeba option

README.md

@@ -42,6 +42,10 @@
             - https://pan.baidu.com/s/1eSNpdRG#list/path=%2FCelebA%2FAnno&parentPath=%2F or
             - https://drive.google.com/drive/folders/0B7EVK8r0v71pOC0wOVZlQnFfaGs
 
+- [HD-Celeba](https://github.com/LynnHo/HD-CelebA-Cropper) (optional)
+    - the images of ***img_align_celeba.zip*** are low resolution and uncropped, higher resolution and cropped images are available [here](https://github.com/LynnHo/HD-CelebA-Cropper)
+    - the high quality data should be placed in ***./data/img_crop_celeba/\*.jpg***
+
 - Examples of training
     - see [examples.md](./examples.md) for more examples
 
@@ -74,6 +78,23 @@
             --experiment_name 384_shortcut1_inject1_none
             ```
 
+        - for 384x384 HD images (need [HD-Celeba](https://github.com/LynnHo/HD-CelebA-Cropper))
+
+            ```console
+            CUDA_VISIBLE_DEVICES=0 \
+            python train.py \
+            --img_size 384 \
+            --enc_dim 48 \
+            --dec_dim 48 \
+            --dis_dim 48 \
+            --dis_fc_dim 512 \
+            --shortcut_layers 1 \
+            --inject_layers 1 \
+            --n_sample 24 \
+            --experiment_name 384_shortcut1_inject1_none \
+            --use_cropped_img
+            ```
+
     - tensorboard for loss visualization
 
         ```console

data.py

@@ -154,14 +154,22 @@ class Celeba(Dataset):
                 'Wearing_Necklace': 37, 'Wearing_Necktie': 38, 'Young': 39}
 
     def __init__(self, data_dir, atts, img_resize, batch_size, prefetch_batch=2, drop_remainder=True,
-                 num_threads=16, shuffle=True, buffer_size=4096, repeat=-1, sess=None, part='train'):
+                 num_threads=16, shuffle=True, buffer_size=4096, repeat=-1, sess=None, part='train', crop=True):
         super(Celeba, self).__init__()
 
         list_file = os.path.join(data_dir, 'list_attr_celeba.txt')
-        img_dir = os.path.join(data_dir, 'img_align_celeba')
+        if crop:
+            img_dir_jpg = os.path.join(data_dir, 'img_align_celeba')
+            img_dir_png = os.path.join(data_dir, 'img_align_celeba_png')
+        else:
+            img_dir_jpg = os.path.join(data_dir, 'img_crop_celeba')
+            img_dir_png = os.path.join(data_dir, 'img_crop_celeba_png')
 
         names = np.loadtxt(list_file, skiprows=2, usecols=[0], dtype=np.str)
-        img_paths = [os.path.join(img_dir, name) for name in names]
+        if os.path.exists(img_dir_png):
+            img_paths = [os.path.join(img_dir_png, name.replace('jpg', 'png')) for name in names]
+        elif os.path.exists(img_dir_jpg):
+            img_paths = [os.path.join(img_dir_jpg, name) for name in names]
 
         att_id = [Celeba.att_dict[att] + 1 for att in atts]
         labels = np.loadtxt(list_file, skiprows=2, usecols=att_id, dtype=np.int64)
@@ -177,12 +185,12 @@ def __init__(self, data_dir, atts, img_resize, batch_size, prefetch_batch=2, dro
             img_size = 170
 
         def _map_func(img, label):
-            img = tf.image.crop_to_bounding_box(img, offset_h, offset_w, img_size, img_size)
-            img = tf.image.resize_images(img, [img_resize, img_resize]) / 127.5 - 1
+            if crop:
+                img = tf.image.crop_to_bounding_box(img, offset_h, offset_w, img_size, img_size)
+            # img = tf.image.resize_images(img, [img_resize, img_resize]) / 127.5 - 1
             # or
-            # img = tf.image.resize_images(img, [img_resize, img_resize], tf.image.ResizeMethod.BICUBIC)
-            # img = (img - tf.reduce_min(img)) / (tf.reduce_max(img) - tf.reduce_min(img))
-            # img = img * 2 - 1
+            img = tf.image.resize_images(img, [img_resize, img_resize], tf.image.ResizeMethod.BICUBIC)
+            img = tf.clip_by_value(img, 0, 255) / 127.5 - 1
             label = (label + 1) // 2
             return img, label
 

test.py

@@ -46,6 +46,7 @@
 thres_int = args['thres_int']
 test_int = args_.test_int
 # others
+use_cropped_img = args['use_cropped_img']
 experiment_name = args_.experiment_name
 
 
@@ -55,7 +56,7 @@
 
 # data
 sess = tl.session()
-te_data = data.Celeba('./data', atts, img_size, 1, part='test', sess=sess)
+te_data = data.Celeba('./data', atts, img_size, 1, part='test', sess=sess, crop=not use_cropped_img)
 
 # models
 Genc = partial(models.Genc, dim=enc_dim, n_layers=enc_layers)

test_multi.py

@@ -48,6 +48,7 @@
 thres_int = args['thres_int']
 test_ints = args_.test_ints
 # others
+use_cropped_img = args['use_cropped_img']
 experiment_name = args_.experiment_name
 
 assert test_atts is not None, 'test_atts should be chosen in %s' % (str(atts))
@@ -63,7 +64,7 @@
 
 # data
 sess = tl.session()
-te_data = data.Celeba('./data', atts, img_size, 1, part='test', sess=sess)
+te_data = data.Celeba('./data', atts, img_size, 1, part='test', sess=sess, crop=not use_cropped_img)
 
 # models
 Genc = partial(models.Genc, dim=enc_dim, n_layers=enc_layers)

test_slide.py

@@ -52,6 +52,7 @@
 test_int_max = args_.test_int_max
 n_slide = args_.n_slide
 # others
+use_cropped_img = args['use_cropped_img']
 experiment_name = args_.experiment_name
 
 assert test_att is not None, 'test_att should be chosen in %s' % (str(atts))
@@ -63,7 +64,7 @@
 
 # data
 sess = tl.session()
-te_data = data.Celeba('./data', atts, img_size, 1, part='test', sess=sess)
+te_data = data.Celeba('./data', atts, img_size, 1, part='test', sess=sess, crop=not use_cropped_img)
 
 # models
 Genc = partial(models.Genc, dim=enc_dim, n_layers=enc_layers)

train.py

@@ -28,26 +28,27 @@
 att_default = ['Bald', 'Bangs', 'Black_Hair', 'Blond_Hair', 'Brown_Hair', 'Bushy_Eyebrows', 'Eyeglasses', 'Male', 'Mouth_Slightly_Open', 'Mustache', 'No_Beard', 'Pale_Skin', 'Young']
 parser.add_argument('--atts', dest='atts', default=att_default, choices=data.Celeba.att_dict.keys(), nargs='+', help='attributes to learn')
 parser.add_argument('--img_size', dest='img_size', type=int, default=128, help='image size')
-parser.add_argument('--shortcut_layers', dest='shortcut_layers', type=int, default=1, help='shortcut_layers')
-parser.add_argument('--inject_layers', dest='inject_layers', type=int, default=0, help='inject_layers')
-parser.add_argument('--enc_dim', dest='enc_dim', type=int, default=64, help='enc_dim')
-parser.add_argument('--dec_dim', dest='dec_dim', type=int, default=64, help='dec_dim')
-parser.add_argument('--dis_dim', dest='dis_dim', type=int, default=64, help='dis_dim')
-parser.add_argument('--dis_fc_dim', dest='dis_fc_dim', type=int, default=1024, help='dis_fc_dim')
-parser.add_argument('--enc_layers', dest='enc_layers', type=int, default=5, help='enc_layers')
-parser.add_argument('--dec_layers', dest='dec_layers', type=int, default=5, help='dec_layers')
-parser.add_argument('--dis_layers', dest='dis_layers', type=int, default=5, help='dis_layers')
+parser.add_argument('--shortcut_layers', dest='shortcut_layers', type=int, default=1)
+parser.add_argument('--inject_layers', dest='inject_layers', type=int, default=0)
+parser.add_argument('--enc_dim', dest='enc_dim', type=int, default=64)
+parser.add_argument('--dec_dim', dest='dec_dim', type=int, default=64)
+parser.add_argument('--dis_dim', dest='dis_dim', type=int, default=64)
+parser.add_argument('--dis_fc_dim', dest='dis_fc_dim', type=int, default=1024)
+parser.add_argument('--enc_layers', dest='enc_layers', type=int, default=5)
+parser.add_argument('--dec_layers', dest='dec_layers', type=int, default=5)
+parser.add_argument('--dis_layers', dest='dis_layers', type=int, default=5)
 # training
 parser.add_argument('--epoch', dest='epoch', type=int, default=200, help='# of epochs')
 parser.add_argument('--batch_size', dest='batch_size', type=int, default=32, help='batch size')
 parser.add_argument('--lr', dest='lr', type=float, default=0.0002, help='learning rate')
 parser.add_argument('--n_d', dest='n_d', type=int, default=5, help='# of d updates per g update')
-parser.add_argument('--b_distribution', dest='b_distribution', default='none', choices=['none', 'uniform', 'truncated_normal'], help='b_distribution')
-parser.add_argument('--thres_int', dest='thres_int', type=float, default=0.5, help='thres_int')
-parser.add_argument('--test_int', dest='test_int', type=float, default=1.0, help='test_int')
+parser.add_argument('--b_distribution', dest='b_distribution', default='none', choices=['none', 'uniform', 'truncated_normal'])
+parser.add_argument('--thres_int', dest='thres_int', type=float, default=0.5)
+parser.add_argument('--test_int', dest='test_int', type=float, default=1.0)
 parser.add_argument('--n_sample', dest='n_sample', type=int, default=64, help='# of sample images')
 # others
-parser.add_argument('--experiment_name', dest='experiment_name', default=datetime.datetime.now().strftime("%I:%M%p on %B %d, %Y"), help='experiment_name')
+parser.add_argument('--use_cropped_img', dest='use_cropped_img', action='store_true')
+parser.add_argument('--experiment_name', dest='experiment_name', default=datetime.datetime.now().strftime("%I:%M%p on %B %d, %Y"))
 
 args = parser.parse_args()
 # model
@@ -73,6 +74,7 @@
 test_int = args.test_int
 n_sample = args.n_sample
 # others
+use_cropped_img = args.use_cropped_img
 experiment_name = args.experiment_name
 
 pylib.mkdir('./output/%s' % experiment_name)
@@ -86,8 +88,8 @@
 
 # data
 sess = tl.session()
-tr_data = data.Celeba('./data', atts, img_size, batch_size, part='train', sess=sess)
-val_data = data.Celeba('./data', atts, img_size, n_sample, part='val', shuffle=False, sess=sess)
+tr_data = data.Celeba('./data', atts, img_size, batch_size, part='train', sess=sess, crop=not use_cropped_img)
+val_data = data.Celeba('./data', atts, img_size, n_sample, part='val', shuffle=False, sess=sess, crop=not use_cropped_img)
 
 # models
 Genc = partial(models.Genc, dim=enc_dim, n_layers=enc_layers)