File size: 6,153 Bytes
9ac69b4
 
 
34bfe14
fcb49fd
20549c6
 
9ac69b4
 
 
 
756e2bd
 
9ac69b4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
31e6d73
9ac69b4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
756e2bd
fcb49fd
 
 
 
20549c6
 
 
 
fcb49fd
 
9ac69b4
756e2bd
9ac69b4
9107020
8b1d9c1
756e2bd
 
8b1d9c1
756e2bd
 
 
 
 
 
8b1d9c1
20549c6
756e2bd
 
 
 
 
 
9ac69b4
756e2bd
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
from huggingface_hub import hf_hub_download


Rain_Princess = hf_hub_download(repo_id="maze/FastStyleTransfer", filename="Rain_Princess_512.pth")
The_Scream = hf_hub_download(repo_id="maze/FastStyleTransfer", filename="Scream_512.pth")
The_Mosaic = hf_hub_download(repo_id="maze/FastStyleTransfer", filename="Mosaic_512.pth")
Starry_Night = hf_hub_download(repo_id="maze/FastStyleTransfer", filename="Starry_Night_512.pth")


import numpy as np
from PIL import Image
import gradio as gr

import torch
import torch.nn as nn

import torchvision.transforms as transforms
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")


class TransformerNetwork(nn.Module):
    def __init__(self, tanh_multiplier=None):
        super(TransformerNetwork, self).__init__()
        self.ConvBlock = nn.Sequential(
            ConvLayer(3, 32, 9, 1),
            nn.ReLU(),
            ConvLayer(32, 64, 3, 2),
            nn.ReLU(),
            ConvLayer(64, 128, 3, 2),
            nn.ReLU()
        )
        self.ResidualBlock = nn.Sequential(
            ResidualLayer(128, 3), 
            ResidualLayer(128, 3), 
            ResidualLayer(128, 3), 
            ResidualLayer(128, 3), 
            ResidualLayer(128, 3)
        )
        self.DeconvBlock = nn.Sequential(
            DeconvLayer(128, 64, 3, 2, 1),
            nn.ReLU(),
            DeconvLayer(64, 32, 3, 2, 1),
            nn.ReLU(),
            ConvLayer(32, 3, 9, 1, norm="None")
        )
        self.tanh_multiplier = tanh_multiplier

    def forward(self, x):
        x = self.ConvBlock(x)
        x = self.ResidualBlock(x)
        x = self.DeconvBlock(x)
        if isinstance(self.tanh_multiplier, int):
            x = self.tanh_multiplier * F.tanh(x)
        return x

class ConvLayer(nn.Module):
    def __init__(self, in_channels, out_channels, kernel_size, stride, norm="instance"):
        super(ConvLayer, self).__init__()
        padding_size = kernel_size // 2
        self.pad = nn.ReflectionPad2d(padding_size)
        self.conv = nn.Conv2d(in_channels, out_channels, kernel_size, stride)
        if norm == "instance":
            self.norm = nn.InstanceNorm2d(out_channels, affine=True)
        elif norm == "batch":
            self.norm = nn.BatchNorm2d(out_channels, affine=True)
        else:
            self.norm = nn.Identity()

    def forward(self, x):
        x = self.pad(x)
        x = self.conv(x)
        x = self.norm(x)
        return x

class ResidualLayer(nn.Module):
    def __init__(self, channels=128, kernel_size=3):
        super(ResidualLayer, self).__init__()
        self.conv1 = ConvLayer(channels, channels, kernel_size, stride=1)
        self.relu = nn.ReLU()
        self.conv2 = ConvLayer(channels, channels, kernel_size, stride=1)

    def forward(self, x):
        identity = x
        out = self.relu(self.conv1(x))
        out = self.conv2(out)
        out = out + identity
        return out

class DeconvLayer(nn.Module):
    def __init__(self, in_channels, out_channels, kernel_size, stride, output_padding, norm="instance"):
        super(DeconvLayer, self).__init__()

        padding_size = kernel_size // 2
        self.conv_transpose = nn.ConvTranspose2d(in_channels, out_channels, kernel_size, stride, padding_size, output_padding)
        if norm == "instance":
            self.norm = nn.InstanceNorm2d(out_channels, affine=True)
        elif norm == "batch":
            self.norm = nn.BatchNorm2d(out_channels, affine=True)
        else:
            self.norm = nn.Identity()

    def forward(self, x):
        x = self.conv_transpose(x)
        out = self.norm(x)
        return out


mean = np.array([0.485, 0.456, 0.406])
std = np.array([0.229, 0.224, 0.225])

transformer = TransformerNetwork().to(device)

transformer.eval()

transform = transforms.Compose([
    transforms.Resize(512),
    transforms.ToTensor(),
    transforms.Normalize(mean, std),
])

denormalize = transforms.Normalize(
   mean= [-m/s for m, s in zip(mean, std)],
   std= [1/s for s in std]
)
tensor2Image = transforms.ToPILImage()

@torch.no_grad()
def process(image, model):
    image = transform(image).to(device)
    image = image.unsqueeze(dim=0)

    image = denormalize(model(image)).cpu()
    image = torch.clamp(image.squeeze(dim=0), 0, 1)
    image = tensor2Image(image)

    return image
    

def main(image, backbone, style):
    if style == "The Scream":
        transformer.load_state_dict(torch.load(The_Scream, map_location=torch.device('cpu')))
    elif style == "Rain Princess":
        transformer.load_state_dict(torch.load(Rain_Princess, map_location=torch.device('cpu')))
    elif style == "The Mosaic":
        transformer.load_state_dict(torch.load(The_Mosaic, map_location=torch.device('cpu')))
    elif style == "Starry Night":
        transformer.load_state_dict(torch.load(Starry_Night, map_location=torch.device('cpu')))
    else:
        transformer.load_state_dict(torch.load(Rain_Princess, map_location=torch.device('cpu')))
    image = Image.fromarray(image)
    isize = image.size
    image = process(image, transformer)
    s = f"The output image {str(image.size)} is processed by {backbone} based on input image {str(isize)}. <br> Please <b>rate</b> the generated image through the <b>Flag</b> button below!"
    print(s)
    return image, s

# "Standard ResNet50", "VGG19"
gr.Interface(
    title = "Stylize", 
    description = "Image generated based on Fast Style Transfer",
    fn = main, 
    inputs = [
        gr.inputs.Image(), 
        gr.inputs.Radio(["Robust ResNet50"], label="Backbone"),
        gr.inputs.Dropdown(["The Scream", "Rain Princess", "Starry Night", "The Mosaic"], type="value", default="Rain Princess", label="style")
    ], 
    outputs = [gr.outputs.Image(label="Stylized"), gr.outputs.HTML(label="Comment")],
    # examples = [
    #     []
    # ],
    # live = True,    # the interface will recalculate as soon as the user input changes.
    allow_flagging = "manual",
    flagging_options = ["Excellect", "Moderate", "Bad"],
    flagging_dir = "flagged",
    allow_screenshot = False,
).launch()
# iface.launch(enable_queue=True, cache_examples=True, debug=True)