Spaces:
Running
Running
Streamlit upload
Browse files
app.py
ADDED
@@ -0,0 +1,111 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import matplotlib.pyplot as plt
|
2 |
+
import numpy as np
|
3 |
+
import streamlit as st
|
4 |
+
import torch
|
5 |
+
from huggingface_hub import PyTorchModelHubMixin
|
6 |
+
from PIL import Image
|
7 |
+
from torchvision import transforms
|
8 |
+
from torchvision.transforms.functional import to_pil_image
|
9 |
+
|
10 |
+
from model import ICN
|
11 |
+
|
12 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
13 |
+
|
14 |
+
|
15 |
+
def mask_processing(x):
|
16 |
+
if x > 90:
|
17 |
+
return 140
|
18 |
+
elif x < 80:
|
19 |
+
return 0
|
20 |
+
else:
|
21 |
+
return 255
|
22 |
+
|
23 |
+
|
24 |
+
def grid_to_heatmap(grid, size=1024):
|
25 |
+
mask = to_pil_image(grid.view(7, 7))
|
26 |
+
mask = mask.resize((size, size), Image.BICUBIC)
|
27 |
+
mask = Image.eval(mask, mask_processing)
|
28 |
+
|
29 |
+
colormap = plt.get_cmap("Wistia")
|
30 |
+
heatmap = np.array(colormap(mask))
|
31 |
+
heatmap = (heatmap * 255).astype(np.uint8)
|
32 |
+
heatmap = Image.fromarray(heatmap)
|
33 |
+
|
34 |
+
return heatmap, mask
|
35 |
+
|
36 |
+
|
37 |
+
def summary_image(img, fake, prediction):
|
38 |
+
prediction -= prediction.min()
|
39 |
+
prediction = prediction / prediction.max()
|
40 |
+
|
41 |
+
size = 1024
|
42 |
+
|
43 |
+
img1 = img.resize((size, size))
|
44 |
+
img2 = fake.resize((size, size))
|
45 |
+
|
46 |
+
heatmap, mask = grid_to_heatmap(prediction)
|
47 |
+
img1.paste(heatmap, (0, 0), mask)
|
48 |
+
img2.paste(heatmap, (0, 0), mask)
|
49 |
+
|
50 |
+
return img1, img2
|
51 |
+
|
52 |
+
|
53 |
+
@st.cache_resource
|
54 |
+
def load_model():
|
55 |
+
model = torch.jit.load("traced_model.pt")
|
56 |
+
model.eval().to(device)
|
57 |
+
return model
|
58 |
+
|
59 |
+
|
60 |
+
model = ICN.from_pretrained("AlexBlck/image-comparator").eval().to(device)
|
61 |
+
|
62 |
+
# model = load_model()
|
63 |
+
|
64 |
+
st.title("Image Comparator Network")
|
65 |
+
|
66 |
+
st.write("## Upload a pair of images")
|
67 |
+
cols = st.columns(2)
|
68 |
+
with cols[0]:
|
69 |
+
im1 = st.file_uploader("Image 1", type=["jpg", "png"])
|
70 |
+
with cols[1]:
|
71 |
+
im2 = st.file_uploader("Image 2", type=["jpg", "png"])
|
72 |
+
|
73 |
+
if not (im1 and im2):
|
74 |
+
st.stop()
|
75 |
+
|
76 |
+
btn = st.button("Run")
|
77 |
+
if not btn:
|
78 |
+
st.stop()
|
79 |
+
|
80 |
+
im1 = Image.open(im1).convert("RGB")
|
81 |
+
im2 = Image.open(im2).convert("RGB")
|
82 |
+
|
83 |
+
tr = transforms.Compose(
|
84 |
+
[
|
85 |
+
transforms.Resize(size=(224, 224)),
|
86 |
+
transforms.ToTensor(),
|
87 |
+
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
|
88 |
+
]
|
89 |
+
)
|
90 |
+
|
91 |
+
img = torch.vstack((tr(im1), tr(im2))).unsqueeze(0)
|
92 |
+
|
93 |
+
heatmap, cl = model(img.to(device))
|
94 |
+
confs = torch.softmax(cl, dim=1)
|
95 |
+
pred = torch.argmax(confs, dim=1).item()
|
96 |
+
|
97 |
+
if pred == 0:
|
98 |
+
st.success("No Manipulation Detected")
|
99 |
+
heatmap *= 0
|
100 |
+
elif pred == 1:
|
101 |
+
st.warning("Manipulation Detected!")
|
102 |
+
else:
|
103 |
+
st.error("Images are not related.")
|
104 |
+
heatmap *= 0
|
105 |
+
|
106 |
+
img1, img2 = summary_image(im1, im2, heatmap[0])
|
107 |
+
cols = st.columns(2)
|
108 |
+
with cols[0]:
|
109 |
+
st.image(img1)
|
110 |
+
with cols[1]:
|
111 |
+
st.image(img2)
|
model.py
ADDED
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import torch
|
2 |
+
import torch.nn.functional as F
|
3 |
+
from huggingface_hub import PyTorchModelHubMixin
|
4 |
+
from torch import nn
|
5 |
+
from torchvision import models
|
6 |
+
|
7 |
+
|
8 |
+
class ICN(nn.Module, PyTorchModelHubMixin):
|
9 |
+
def __init__(self):
|
10 |
+
super().__init__()
|
11 |
+
|
12 |
+
cnn = models.resnet50(pretrained=False)
|
13 |
+
self.cnn_head = nn.Sequential(
|
14 |
+
*list(cnn.children())[:4],
|
15 |
+
*list(list(list(cnn.children())[4].children())[0].children())[:4],
|
16 |
+
)
|
17 |
+
self.cnn_tail = nn.Sequential(
|
18 |
+
*list(list(cnn.children())[4].children()
|
19 |
+
)[1:], *list(cnn.children())[5:-2]
|
20 |
+
)
|
21 |
+
|
22 |
+
self.conv1 = nn.Conv2d(128, 256, 3, padding=1)
|
23 |
+
self.bn1 = nn.BatchNorm2d(num_features=256)
|
24 |
+
|
25 |
+
self.fc1 = nn.Linear(2048 * 7 * 7, 256)
|
26 |
+
self.fc2 = nn.Linear(256, 7 * 7)
|
27 |
+
|
28 |
+
self.cls_fc = nn.Linear(256, 3)
|
29 |
+
|
30 |
+
self.criterion = nn.CrossEntropyLoss()
|
31 |
+
|
32 |
+
def forward(self, x):
|
33 |
+
# Input: [-1, 6, 224, 224]
|
34 |
+
real = x[:, :3, :, :]
|
35 |
+
fake = x[:, 3:, :, :]
|
36 |
+
|
37 |
+
# Push both images through pretrained backbone
|
38 |
+
real_features = F.relu(self.cnn_head(real)) # [-1, 64, 56, 56]
|
39 |
+
fake_features = F.relu(self.cnn_head(fake)) # [-1, 64, 56, 56]
|
40 |
+
|
41 |
+
# [-1, 128, 56, 56]
|
42 |
+
combined = torch.cat((real_features, fake_features), 1)
|
43 |
+
|
44 |
+
x = self.conv1(combined) # [-1, 256, 56, 56]
|
45 |
+
x = self.bn1(x)
|
46 |
+
x = F.relu(x)
|
47 |
+
|
48 |
+
x = self.cnn_tail(x)
|
49 |
+
x = x.view(-1, 2048 * 7 * 7)
|
50 |
+
|
51 |
+
# Final feature [-1, 256]
|
52 |
+
d = F.relu(self.fc1(x))
|
53 |
+
|
54 |
+
# Heatmap [-1, 49]
|
55 |
+
grid = self.fc2(d)
|
56 |
+
|
57 |
+
# Classifier [-1, 1]
|
58 |
+
cl = self.cls_fc(d)
|
59 |
+
|
60 |
+
return grid, cl
|