Spaces:
Running
Running
add new func
Browse files- models/SRFlow/srflow.py +27 -32
models/SRFlow/srflow.py
CHANGED
@@ -4,9 +4,10 @@ import sys
|
|
4 |
sys.path.append('models')
|
5 |
from SRFlow.code import imread, impad, load_model, t, rgb
|
6 |
from PIL import Image
|
7 |
-
|
|
|
8 |
|
9 |
-
def return_SRFlow_result(lr, conf_path='models/SRFlow/code/confs/SRFlow_DF2K_4X.yml'
|
10 |
"""
|
11 |
Apply Super-Resolution using SRFlow model to the input LR (low-resolution) image.
|
12 |
|
@@ -39,43 +40,37 @@ def return_SRFlow_result(lr, conf_path='models/SRFlow/code/confs/SRFlow_DF2K_4X.
|
|
39 |
sr = Image.fromarray((sr).astype('uint8'))
|
40 |
return sr
|
41 |
|
42 |
-
def
|
43 |
"""
|
44 |
-
Apply Super-Resolution using SRFlow model to the input
|
45 |
|
46 |
Args:
|
47 |
-
-
|
48 |
-
- conf_path (str): Configuration file path for the SRFlow model. Default is SRFlow_DF2K_4X.yml.
|
49 |
-
- heat (float): Heat parameter for the SRFlow model. Default is 0.6.
|
50 |
|
51 |
Returns:
|
52 |
-
-
|
53 |
"""
|
54 |
-
|
|
|
55 |
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
lr = impad(lr, bottom=int(np.ceil(h / pad_factor) * pad_factor - h),
|
62 |
-
right=int(np.ceil(w / pad_factor) * pad_factor - w))
|
63 |
-
|
64 |
-
lr_t = t(lr)
|
65 |
-
heat = opt['heat']
|
66 |
-
|
67 |
-
sr_t = model.get_sr(lq=lr_t, heat=heat)
|
68 |
|
69 |
-
|
70 |
-
|
71 |
-
sr = sr.unsqueeze(0).permute(0, 3, 1, 2)
|
72 |
-
|
73 |
-
return sr
|
74 |
|
75 |
if __name__ == '__main__':
|
76 |
-
|
77 |
-
|
78 |
-
|
79 |
-
|
80 |
-
|
81 |
-
print(sr.
|
|
|
|
|
|
|
|
|
|
|
|
4 |
sys.path.append('models')
|
5 |
from SRFlow.code import imread, impad, load_model, t, rgb
|
6 |
from PIL import Image
|
7 |
+
import matplotlib.pyplot as plt
|
8 |
+
from torchvision.transforms import PILToTensor, ToPILImage
|
9 |
|
10 |
+
def return_SRFlow_result(lr, conf_path='models/SRFlow/code/confs/SRFlow_DF2K_4X.yml'):
|
11 |
"""
|
12 |
Apply Super-Resolution using SRFlow model to the input LR (low-resolution) image.
|
13 |
|
|
|
40 |
sr = Image.fromarray((sr).astype('uint8'))
|
41 |
return sr
|
42 |
|
43 |
+
def return_SRFlow_result_from_tensor(lr_tensor):
|
44 |
"""
|
45 |
+
Apply Super-Resolution using SRFlow model to the input batched BCHW tensor.
|
46 |
|
47 |
Args:
|
48 |
+
- lr_tensor: Batched BCHW tensor
|
|
|
|
|
49 |
|
50 |
Returns:
|
51 |
+
- sr_tensor: Processed batched BCHW tensor
|
52 |
"""
|
53 |
+
batch_size = lr_tensor.shape[0]
|
54 |
+
sr_list = []
|
55 |
|
56 |
+
for b in range(batch_size):
|
57 |
+
lr_image = ToPILImage()(lr_tensor[b])
|
58 |
+
sr_image = return_SRFlow_result(lr_image)
|
59 |
+
sr_tensor = PILToTensor()(sr_image).unsqueeze(0)
|
60 |
+
sr_list.append(sr_tensor)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
61 |
|
62 |
+
sr_tensor = torch.cat(sr_list, dim=0)
|
63 |
+
return sr_tensor
|
|
|
|
|
|
|
64 |
|
65 |
if __name__ == '__main__':
|
66 |
+
lr = Image.open('images/demo.png')
|
67 |
+
|
68 |
+
lr_tensor = PILToTensor()(lr).unsqueeze(0)
|
69 |
+
|
70 |
+
sr = return_SRFlow_result_from_tensor(lr_tensor)
|
71 |
+
print(sr.shape)
|
72 |
+
|
73 |
+
plt.imshow(np.transpose(sr[0].cpu().detach().numpy(), (1, 2, 0)))
|
74 |
+
plt.axis('off')
|
75 |
+
plt.title('Super-Resolved Image')
|
76 |
+
plt.show()
|