File size: 5,828 Bytes
7e08bf1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
604f77c
7e08bf1
604f77c
7e08bf1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c077072
 
 
 
 
 
 
 
7e08bf1
 
 
604f77c
7e08bf1
c077072
7e08bf1
 
c077072
 
 
 
 
604f77c
7e08bf1
 
604f77c
7e08bf1
144ffb8
7e08bf1
604f77c
 
 
7e08bf1
 
144ffb8
 
 
 
 
 
c077072
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7e08bf1
 
 
c077072
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7e08bf1
c077072
 
 
 
 
 
7e08bf1
 
 
ac42c22
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
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
import gradio as gr
import torch
import torchvision
from PIL import Image
import numpy as np
import matplotlib.pyplot as plt
import matplotlib
matplotlib.use('Agg')
import io
import os

# Import your models
from models.feature_extractor import FeatureExtractor, FeatureExtractorDepth
from models.projector import SiameseProjector
from models.fuser import DoubleCrossAttentionFusion
from loaders.loader_utils import SquarePad

# Configuration
CHECKPOINT_PATH = './checkpoints'
MODEL_LABEL = 'multimodal_15k_10inp'
EPOCHS = 120
BATCH_SIZE = 4
IMAGE_SIZE = 896

# Load models
print("Loading models...")
device = "cuda" if torch.cuda.is_available() else "cpu"
print(f"Using device: {device}")

model_name = f'{MODEL_LABEL}_{EPOCHS}ep_{BATCH_SIZE}bs'

rgb_transform = torchvision.transforms.Compose([
    SquarePad(),
    torchvision.transforms.Resize((IMAGE_SIZE, IMAGE_SIZE), 
                                  interpolation=torchvision.transforms.InterpolationMode.BICUBIC),
    torchvision.transforms.ToTensor(),
    torchvision.transforms.Normalize(mean=[0.485, 0.456, 0.406], 
                                    std=[0.229, 0.224, 0.225]),
    torchvision.transforms.Lambda(lambda img: img.unsqueeze(0)),
])

fe_rgb = FeatureExtractor().to(device).eval()
fe_depth = FeatureExtractorDepth().to(device).eval()

fusion_block = DoubleCrossAttentionFusion(hidden_dim=fe_rgb.embed_dim).to(device)
fusion_block.load_state_dict(torch.load(
    os.path.join(CHECKPOINT_PATH, f'fusion_block_{model_name}.pth'), 
    weights_only=False,
    map_location=device
))
fusion_block.eval()

projector = SiameseProjector(inner_features=fe_rgb.embed_dim).to(device)
projector.load_state_dict(torch.load(
    os.path.join(CHECKPOINT_PATH, f'projector_{model_name}.pth'), 
    weights_only=False,
    map_location=device
))
projector.eval()

print("Models loaded successfully!")

def detect_manipulation(image):
    """Process image and return heatmap"""
    if image is None:
        return None
    
    # Convert to PIL
    if isinstance(image, np.ndarray):
        rgb_input = Image.fromarray(image.astype('uint8')).convert('RGB')
    else:
        rgb_input = image.convert('RGB')
    
    original_size = rgb_input.size
    
    # Transform and process
    rgb = rgb_transform(rgb_input)
    rgb = rgb.to(device)
    
    with torch.no_grad():
        rgb_feat = fe_rgb(rgb)
        depth_feat = fe_depth(rgb)
        fused_feat = fusion_block(rgb_feat, depth_feat)
        _, segmentation_map = projector(fused_feat)
        segmentation_map = torch.sigmoid(segmentation_map)
        
        # Resize back to original
        segmentation_map = torch.nn.functional.interpolate(
            segmentation_map, 
            size=[max(original_size), max(original_size)], 
            mode='bilinear'
        ).squeeze()
        segmentation_map = torchvision.transforms.functional.center_crop(
            segmentation_map, 
            original_size[::-1]
        )
        
        heatmap = segmentation_map.cpu().detach().numpy()
    
    # Create visualization with exact size
    # Calculate figure size to match image dimensions
    dpi = 100
    fig_height = original_size[1] / dpi
    fig_width = original_size[0] / dpi
    
    fig = plt.figure(figsize=(fig_width, fig_height), dpi=dpi)
    ax = fig.add_axes([0, 0, 1, 1])  # No margins
    ax.imshow(heatmap, cmap='jet')
    ax.axis('off')
    
    # Convert to numpy array
    buf = io.BytesIO()
    plt.savefig(buf, format='png', bbox_inches='tight', pad_inches=0, dpi=dpi)
    buf.seek(0)
    result_image = Image.open(buf)
    
    # Ensure exact size match by resizing if needed
    if result_image.size != original_size:
        result_image = result_image.resize(original_size, Image.LANCZOS)
    
    result_array = np.array(result_image)
    plt.close(fig)
    
    return result_array

# Custom CSS for styling
custom_css = """
.gradio-container {
    font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, sans-serif;
}
#title {
    text-align: center;
    font-size: 2.5em;
    font-weight: bold;
    margin-bottom: 0.5em;
    background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
    -webkit-background-clip: text;
    -webkit-text-fill-color: transparent;
    background-clip: text;
}
#subtitle {
    text-align: center;
    font-size: 1.2em;
    color: #666;
    margin-bottom: 1em;
}
#info {
    background: #e8f4fd;
    border-left: 4px solid #2196F3;
    padding: 15px;
    border-radius: 5px;
    margin-bottom: 20px;
    color: #1976D2;
}
"""

# Create interface using Gradio 4.x Blocks
with gr.Blocks(css=custom_css, title="RADAR - Image Manipulation Detection") as demo:
    gr.HTML('<h1 id="title">🎯 RADAR</h1>')
    gr.HTML('<p id="subtitle">ReliAble iDentification of inpainted AReas</p>')
    
    gr.HTML('''
        <div id="info">
            <strong>ℹ️ About RADAR:</strong> Upload an image to detect and localize regions 
            that have been manipulated using diffusion-based inpainting models. 
            The output shows a heatmap where red areas indicate detected manipulations.
        </div>
    ''')
    
    with gr.Row():
        with gr.Column():
            input_image = gr.Image(label="Upload Image", type="numpy")
        with gr.Column():
            output_image = gr.Image(label="Manipulation Heatmap", type="numpy")
    
    submit_btn = gr.Button("🔍 Detect Manipulations", variant="primary")
    
    # Connect the button
    submit_btn.click(
        fn=detect_manipulation,
        inputs=input_image,
        outputs=output_image
    )
    
    # Also trigger on image upload
    input_image.change(
        fn=detect_manipulation,
        inputs=input_image,
        outputs=output_image
    )

# Launch
if __name__ == "__main__":
    demo.launch(server_name="0.0.0.0", server_port=7860)