Update app.py

app.py

@@ -7,9 +7,25 @@ from src.predict import process_single_image
 import sys
 sys.path.insert(0, "./src")
 
-def get_example_images(folder_path="/home/xxw/Loupe/ffhq"):
-    return [os.path.join(folder_path, f) for f in os.listdir(folder_path) 
-            if f.lower().endswith(('.png', '.jpg', '.jpeg'))]
+# 自定义主题 - 炫彩现代化
+custom_theme = gr.themes.Default(
+    primary_hue="purple",
+    secondary_hue="pink",
+    neutral_hue="slate",
+    font=[gr.themes.GoogleFont("Poppins"), gr.themes.GoogleFont("Inter"), "Arial", "sans-serif"]
+).set(
+    button_primary_background_fill="linear-gradient(45deg, #667eea 0%, #764ba2 100%)",
+    button_primary_background_fill_hover="linear-gradient(45deg, #764ba2 0%, #667eea 100%)",
+    button_primary_text_color="white",
+    button_secondary_background_fill="linear-gradient(45deg, #f093fb 0%, #f5576c 100%)",
+    button_secondary_background_fill_hover="linear-gradient(45deg, #f5576c 0%, #f093fb 100%)",
+    button_secondary_text_color="white"
+)
+
+def get_example_images(folder_path="ffhq"):
+    """获取示例图片列表"""
+    return sorted([os.path.join(folder_path, f) for f in os.listdir(folder_path) 
+            if f.lower().endswith(('.png', '.jpg', '.jpeg'))])
 
 def safe_extract_prob(cls_probs):
     """安全地从cls_probs中提取概率值"""
@@ -26,74 +42,420 @@ def safe_extract_prob(cls_probs):
         print(f"Error extracting probability: {e}")
         return 0.0
 
-with gr.Blocks(title="Loupe图像伪造检测系统", theme=gr.themes.Soft()) as demo:
-    gr.Markdown("""
-    # Loupe🕵️‍♂️ 图像伪造检测系统
-    ### 上传图像或从示例中选择，系统将检测图像中的伪造区域
+# 创建主界面
+with gr.Blocks(
+    title="Loupe - AI图像伪造检测系统",
+    theme=custom_theme,
+    css="""
+    /* 全局样式 */
+    body {
+        background: linear-gradient(-45deg, #ee7752, #e73c7e, #23a6d5, #23d5ab);
+        background-size: 400% 400%;
+        animation: gradientBG 15s ease infinite;
+        min-height: 100vh;
+    }
+    
+    @keyframes gradientBG {
+        0% { background-position: 0% 50%; }
+        50% { background-position: 100% 50%; }
+        100% { background-position: 0% 50%; }
+    }
+    
+    /* 主容器样式 */
+    .gradio-container {
+        background: rgba(255, 255, 255, 0.95);
+        backdrop-filter: blur(10px);
+        border-radius: 20px;
+        box-shadow: 0 20px 40px rgba(0, 0, 0, 0.1);
+        margin: 20px;
+        padding: 20px;
+    }
+    
+    /* 标题样式 */
+    .title-box {
+        background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
+        padding: 30px;
+        border-radius: 15px;
+        margin-bottom: 30px;
+        box-shadow: 0 15px 35px rgba(102, 126, 234, 0.3);
+        position: relative;
+        overflow: hidden;
+    }
+    
+    .title-box::before {
+        content: '';
+        position: absolute;
+        top: -50%;
+        left: -50%;
+        width: 200%;
+        height: 200%;
+        background: linear-gradient(45deg, transparent, rgba(255, 255, 255, 0.1), transparent);
+        animation: shine 3s infinite;
+    }
+    
+    @keyframes shine {
+        0% { transform: translateX(-100%) translateY(-100%) rotate(45deg); }
+        100% { transform: translateX(100%) translateY(100%) rotate(45deg); }
+    }
+    
+    .title-text {
+        font-weight: 700;
+        font-size: 32px;
+        color: white;
+        margin-bottom: 8px;
+        text-shadow: 2px 2px 4px rgba(0, 0, 0, 0.3);
+        background: linear-gradient(45deg, #fff, #f0f8ff);
+        -webkit-background-clip: text;
+        -webkit-text-fill-color: transparent;
+        background-clip: text;
+    }
+    
+    .subtitle-text {
+        color: rgba(255, 255, 255, 0.9);
+        font-size: 18px;
+        font-weight: 300;
+        text-shadow: 1px 1px 2px rgba(0, 0, 0, 0.2);
+    }
+    
+    /* 输入和结果框样式 */
+    .input-box, .result-box {
+        background: linear-gradient(145deg, rgba(255, 255, 255, 0.9), rgba(248, 250, 252, 0.9));
+        padding: 25px;
+        border-radius: 15px;
+        margin-bottom: 20px;
+        border: 1px solid rgba(255, 255, 255, 0.3);
+        box-shadow: 0 10px 30px rgba(0, 0, 0, 0.1);
+        backdrop-filter: blur(10px);
+        transition: all 0.3s ease;
+    }
+    
+    .input-box:hover, .result-box:hover {
+        transform: translateY(-5px);
+        box-shadow: 0 20px 40px rgba(0, 0, 0, 0.15);
+    }
+    
+    .input-title, .result-title {
+        font-weight: 700;
+        background: linear-gradient(45deg, #667eea, #764ba2);
+        -webkit-background-clip: text;
+        -webkit-text-fill-color: transparent;
+        background-clip: text;
+        margin-bottom: 15px;
+        font-size: 20px;
+        text-shadow: 0 2px 4px rgba(0, 0, 0, 0.1);
+    }
+    
+    /* 按钮样式 */
+    .btn-primary {
+        background: linear-gradient(45deg, #667eea 0%, #764ba2 100%);
+        border: none;
+        border-radius: 25px;
+        padding: 12px 30px;
+        font-weight: 600;
+        text-transform: uppercase;
+        letter-spacing: 1px;
+        box-shadow: 0 10px 20px rgba(102, 126, 234, 0.3);
+        transition: all 0.3s ease;
+    }
+    
+    .btn-primary:hover {
+        transform: translateY(-3px);
+        box-shadow: 0 15px 30px rgba(102, 126, 234, 0.4);
+        background: linear-gradient(45deg, #764ba2 0%, #667eea 100%);
+    }
+    
+    .btn-secondary {
+        background: linear-gradient(45deg, #f093fb 0%, #f5576c 100%);
+        border: none;
+        border-radius: 25px;
+        padding: 10px 25px;
+        font-weight: 600;
+        box-shadow: 0 8px 16px rgba(240, 147, 251, 0.3);
+        transition: all 0.3s ease;
+    }
+    
+    .btn-secondary:hover {
+        transform: translateY(-2px);
+        box-shadow: 0 12px 24px rgba(240, 147, 251, 0.4);
+    }
+    
+    /* 图片上传区域 */
+    #upload_image {
+        min-height: 350px;
+        border: 3px dashed rgba(102, 126, 234, 0.3);
+        border-radius: 15px;
+        background: linear-gradient(45deg, rgba(102, 126, 234, 0.05), rgba(118, 75, 162, 0.05));
+        transition: all 0.3s ease;
+    }
+    
+    #upload_image:hover {
+        border-color: rgba(102, 126, 234, 0.6);
+        background: linear-gradient(45deg, rgba(102, 126, 234, 0.1), rgba(118, 75, 162, 0.1));
+        transform: scale(1.02);
+    }
+    
+    /* 概率显示 */
+    #probability input {
+        font-weight: bold;
+        background: linear-gradient(45deg, #667eea, #764ba2);
+        -webkit-background-clip: text;
+        -webkit-text-fill-color: transparent;
+        background-clip: text;
+        font-size: 1.2em;
+    }
+    
+    #result_text input {
+        font-size: 1.1em;
+        font-weight: 600;
+        background: linear-gradient(45deg, rgba(102, 126, 234, 0.1), rgba(118, 75, 162, 0.1));
+        border-radius: 10px;
+        border: 2px solid rgba(102, 126, 234, 0.2);
+    }
+    
+    /* 画廊样式 */
+    .gallery-item {
+        border-radius: 12px !important;
+        transition: all 0.3s ease;
+        box-shadow: 0 5px 15px rgba(0, 0, 0, 0.1);
+    }
+    
+    .gallery-item:hover {
+        transform: scale(1.05);
+        box-shadow: 0 10px 25px rgba(0, 0, 0, 0.2);
+    }
+    
+    /* 示例按钮 */
+    .example-btn {
+        margin-top: 15px;
+        width: 100%;
+        background: linear-gradient(45deg, #23a6d5 0%, #23d5ab 100%);
+        border-radius: 20px;
+        font-weight: 600;
+        box-shadow: 0 8px 16px rgba(35, 166, 213, 0.3);
+        transition: all 0.3s ease;
+    }
+    
+    .example-btn:hover {
+        transform: translateY(-2px);
+        box-shadow: 0 12px 24px rgba(35, 166, 213, 0.4);
+    }
+    
+    /* Tab 样式 */
+    .tab-nav button {
+        border-radius: 15px 15px 0 0;
+        background: linear-gradient(45deg, rgba(102, 126, 234, 0.8), rgba(118, 75, 162, 0.8));
+        color: white;
+        font-weight: 600;
+        transition: all 0.3s ease;
+    }
+    
+    .tab-nav button:hover {
+        background: linear-gradient(45deg, rgba(118, 75, 162, 0.9), rgba(102, 126, 234, 0.9));
+        transform: translateY(-2px);
+    }
+    
+    /* 滑块样式 */
+    .gr-slider input[type="range"] {
+        background: linear-gradient(45deg, #667eea, #764ba2);
+        border-radius: 10px;
+    }
+    
+    /* 手风琴样式 */
+    .gr-accordion {
+        background: linear-gradient(145deg, rgba(255, 255, 255, 0.8), rgba(248, 250, 252, 0.8));
+        border-radius: 15px;
+        border: 1px solid rgba(102, 126, 234, 0.2);
+        box-shadow: 0 5px 15px rgba(0, 0, 0, 0.1);
+    }
+    
+    /* 炫彩加载动画 */
+    @keyframes rainbow {
+        0% { background-position: 0% 50%; }
+        50% { background-position: 100% 50%; }
+        100% { background-position: 0% 50%; }
+    }
+    
+    .processing {
+        background: linear-gradient(-45deg, #ee7752, #e73c7e, #23a6d5, #23d5ab);
+        background-size: 400% 400%;
+        animation: rainbow 2s ease infinite;
+    }
+    
+    /* 响应式设计 */
+    @media (max-width: 768px) {
+        .title-text { font-size: 24px; }
+        .subtitle-text { font-size: 16px; }
+        .input-box, .result-box { padding: 20px; }
+    }
+    """
+) as demo:
+    
+    # 标题部分 - 炫彩渐变设计
+    with gr.Column(elem_classes="title-box"):
+        gr.Markdown("""
+        <div class="title-text">🔍 Loupe 图像伪造检测系统</div>
+        <div class="subtitle-text">✨ 基于深度学习的图像伪造检测与定位技术</div>
+        """)
+    
+    # 添加装饰性分割线
+    gr.HTML("""
+    <div style="height: 4px; background: linear-gradient(90deg, #667eea, #764ba2, #f093fb, #f5576c, #23a6d5, #23d5ab); 
+                border-radius: 2px; margin: 20px 0; box-shadow: 0 2px 10px rgba(0,0,0,0.2);"></div>
     """)
     
-    with gr.Row():
-        with gr.Column(scale=1):
-            with gr.Tab("上传图像"):
-                image_input = gr.Image(type="pil", label="原始图像")
-                upload_button = gr.Button("检测伪造", variant="primary")
-            
-            with gr.Tab("选择示例"):
-                example_images = get_example_images()
-                example_dropdown = gr.Dropdown(
-                    choices=example_images,
-                    label="选择示例图像",
-                    value=example_images[0] if example_images else None
-                )
-                example_button = gr.Button("检测示例", variant="secondary")
-            
-            with gr.Accordion("高级选项", open=False):
-                threshold = gr.Slider(0, 1, value=0.5, label="检测阈值")
+    # 主界面组件
+    with gr.Row(equal_height=True):
+        with gr.Column(scale=1, min_width=300):
+            # 输入图像区域 - 炫彩设计
+            with gr.Column(elem_classes="input-box"):
+                gr.Markdown("""<div class="input-title">🎨 输入图像</div>""")
+                with gr.Tabs():
+                    with gr.Tab("📤 上传图片", id="upload_tab"):
+                        image_input = gr.Image(type="pil", label="", elem_id="upload_image")
+                        upload_button = gr.Button("🚀 开始检测", variant="primary", size="lg", elem_classes="btn-primary")
+                    
+                    with gr.Tab("🖼️ 示例图片", id="example_tab"):
+                        example_images = get_example_images()
+                        example_gallery = gr.Gallery(
+                            value=example_images,
+                            label="",
+                            columns=4,
+                            rows=None,
+                            height="auto",
+                            object_fit="contain",
+                            allow_preview=True,
+                            selected_index=None
+                        )
+                        # 添加炫彩检测按钮
+                        example_button = gr.Button(
+                            "✨ 检测选中的示例图片", 
+                            variant="primary", 
+                            elem_classes="example-btn"
+                        )
+                        # 隐藏组件用于存储选中索引
+                        selected_index = gr.Number(visible=False)
+                
+            with gr.Accordion("⚙️ 高级设置", open=False):
+                threshold = gr.Slider(0, 1, value=0.5, step=0.01, label="🎯 检测敏感度")
+                gr.HTML("""
+                <div style="background: linear-gradient(45deg, rgba(102,126,234,0.1), rgba(118,75,162,0.1)); 
+                           padding: 10px; border-radius: 8px; margin-top: 10px;">
+                    <small style="color: #667eea; font-weight: 500;">💡 调整数值可改变检测的严格程度</small>
+                </div>
+                """)
         
-        with gr.Column(scale=1):
-            gr.Markdown("### 检测结果")
-            with gr.Tabs():
-                with gr.Tab("处理后的图像"):
-                    output_image = gr.Image(label="伪造检测结果", interactive=False)
+        with gr.Column(scale=1.5, min_width=500):
+            # 检测结果区域 - 炫彩设计
+            with gr.Column(elem_classes="result-box"):
+                gr.Markdown("""<div class="result-title">🎯 检测结果</div>""")
+                with gr.Tabs():
+                    with gr.Tab("🔍 检测效果", id="result_tab"):
+                        output_image = gr.Image(label="伪造区域标记", interactive=False)
+                    
+                    with gr.Tab("⚖️ 对比视图", id="compare_tab"):
+                        with gr.Row():
+                            original_display = gr.Image(label="原始图像", interactive=False)
+                            processed_display = gr.Image(label="检测结果", interactive=False)
                 
-                with gr.Tab("对比视图"):
+                with gr.Group():
                     with gr.Row():
-                        original_display = gr.Image(label="原始图像", interactive=False)
-                        processed_display = gr.Image(label="处理后图像", interactive=False)
-            
-            with gr.Group():
+                        fake_prob = gr.Number(label="🎲 伪造概率", precision=2, elem_id="probability")
+                        result_text = gr.Textbox(label="📝 检测结论", interactive=False, elem_id="result_text")
+                
                 with gr.Row():
-                    fake_prob = gr.Number(label="伪造概率", precision=4)
-                    # Simplified to just show the probability as text
-                    result_text = gr.Textbox(label="检测结果", interactive=False)
-            
-            save_button = gr.Button("保存结果", variant="secondary")
+                    save_button = gr.Button("💾 保存结果", variant="secondary", elem_classes="btn-secondary")
+                    clear_button = gr.Button("🧹 清除", variant="secondary", elem_classes="btn-secondary")
+    
+    # 关于部分 - 炫彩设计
+    with gr.Accordion("🌟 关于系统", open=False):
+        gr.HTML("""
+        <div style="background: linear-gradient(135deg, rgba(102,126,234,0.1), rgba(118,75,162,0.1), rgba(240,147,251,0.1)); 
+                   padding: 20px; border-radius: 15px; border: 1px solid rgba(102,126,234,0.2);">
+            <h3 style="background: linear-gradient(45deg, #667eea, #764ba2); -webkit-background-clip: text; 
+                      -webkit-text-fill-color: transparent; margin-bottom: 15px;">
+                ✨ Loupe 伪造图像检测系统
+            </h3>
+            <div style="display: grid; grid-template-columns: repeat(auto-fit, minmax(200px, 1fr)); gap: 15px;">
+                <div style="background: rgba(102,126,234,0.1); padding: 15px; border-radius: 10px;">
+                    <strong style="color: #667eea;">🚀 技术</strong><br>
+                    基于深度学习的图像伪造检测与定位
+                </div>
+                <div style="background: rgba(118,75,162,0.1); padding: 15px; border-radius: 10px;">
+                    <strong style="color: #764ba2;">⭐ 特点</strong><br>
+                    高精度、实时处理、可解释性强
+                </div>
+                <div style="background: rgba(240,147,251,0.1); padding: 15px; border-radius: 10px;">
+                    <strong style="color: #f093fb;">📱 版本</strong><br>
+                    v2.0.0 炫彩版
+                </div>
+                <div style="background: rgba(245,87,108,0.1); padding: 15px; border-radius: 10px;">
+                    <strong style="color: #f5576c;">👥 开发者</strong><br>
+                    EVOL Lab (jyc, xxw)
+                </div>
+            </div>
+            <div style="margin-top: 20px; padding: 15px; background: linear-gradient(45deg, rgba(35,166,213,0.1), rgba(35,213,171,0.1)); 
+                       border-radius: 10px; border-left: 4px solid #23a6d5;">
+                <strong style="color: #23a6d5;">💡 系统介绍</strong><br>
+                本系统可检测多种图像篡改痕迹，包括复制-移动、拼接、擦除等操作。采用最新的深度学习算法，提供高精度的检测结果和直观的可视化分析。
+            </div>
+        </div>
+        """)
     
-    gr.Markdown("""
-    ---
-    ### 关于
-    - **技术**: Forgery Image Detection and Localization.
-    - **版本**: 1.0.0
+    # 页脚 - 炫彩设计
+    gr.HTML("""
+    <div style="margin-top: 40px; padding: 20px; text-align: center; 
+               background: linear-gradient(135deg, rgba(102,126,234,0.1), rgba(118,75,162,0.1)); 
+               border-radius: 15px; border-top: 2px solid rgba(102,126,234,0.3);">
+        <div style="background: linear-gradient(45deg, #667eea, #764ba2); -webkit-background-clip: text; 
+                   -webkit-text-fill-color: transparent; font-weight: 600; margin-bottom: 10px;">
+            ✨ 感谢使用 Loupe 图像伪造检测系统 ✨
+        </div>
+        <div style="color: #64748b; font-size: 14px;">
+            © 2025 EVOL Lab | 让AI守护图像真实性 🛡️
+        </div>
+        <div style="margin-top: 10px;">
+            <span style="background: linear-gradient(45deg, #f093fb, #f5576c); -webkit-background-clip: text; 
+                        -webkit-text-fill-color: transparent; font-weight: 500;">
+                🌟 科技点亮未来，智能守护真实 🌟
+            </span>
+        </div>
+    </div>
     """)
 
     def process_image(image, threshold_value):
+        """处理上传的图像"""
+        if image is None:
+            return {
+                output_image: None, 
+                fake_prob: 0.0, 
+                result_text: "❌ 请上传有效图像"
+            }
+        
         with tempfile.NamedTemporaryFile(suffix=".png", delete=False) as tmp_file:
             image_path = tmp_file.name
             image.save(image_path)
 
         try:
             processed_img, cls_probs = process_single_image(image_path)
-            
-            # 安全提取概率值
             prob = safe_extract_prob(cls_probs)
-            print(f"Classification probability: {prob:.4f}" if prob is not None else "No cls output")
+            
+            # 根据概率生成炫彩结论
+            if prob > threshold_value + 0.2:
+                conclusion = "🚨 高度疑似伪造"
+                emoji = "🔴"
+            elif prob > threshold_value:
+                conclusion = "⚠️ 可能伪造"
+                emoji = "🟡"
+            else:
+                conclusion = "✅ 未检测到伪造"
+                emoji = "🟢"
             
             return {
                 output_image: processed_img,
                 original_display: image,
                 processed_display: processed_img,
                 fake_prob: prob,
-                result_text: f"伪造概率: {prob:.4f}"
+                result_text: f"{emoji} {conclusion} (概率: {prob:.2f})"
             }
         except Exception as e:
             print(f"Error in processing: {e}")
@@ -102,29 +464,62 @@ with gr.Blocks(title="Loupe图像伪造检测系统", theme=gr.themes.Soft()) as
                 original_display: image,
                 processed_display: None,
                 fake_prob: 0.0,
-                result_text: "处理错误"
+                result_text: f"❌ 处理错误: {str(e)}"
             }
         finally:
             if os.path.exists(image_path):
                 os.unlink(image_path)
-    
-    def process_example(image_path, threshold_value):
+
+    def process_example(example_data, selected_idx, threshold_value):
+        """处理示例图像"""
+        if not example_data or selected_idx is None:
+            return {
+                image_input: None,
+                output_image: None,
+                original_display: None,
+                processed_display: None,
+                fake_prob: 0.0,
+                result_text: "⚠️ 请先选择示例图片",
+                threshold: threshold_value
+            }
+            
         try:
-            processed_img, cls_probs = process_single_image(image_path)
+            selected_idx = int(selected_idx)
+            image_info = example_data[selected_idx]
             
-            # 安全提取概率值
-            prob = safe_extract_prob(cls_probs)
-            print(f"Classification probability: {prob:.4f}" if prob is not None else "No cls output")
+            # 处理不同的数据格式
+            if isinstance(image_info, (tuple, list)):
+                image_path = image_info[0]  # (path, caption)格式
+            elif isinstance(image_info, dict):
+                image_path = image_info.get("name", image_info.get("path"))
+            else:
+                image_path = image_info
+                
+            print(f"Processing selected image (index {selected_idx}): {image_path}")  # 调试日志
             
+            # 处理图像
+            processed_img, cls_probs = process_single_image(image_path)
+            prob = safe_extract_prob(cls_probs)
             original_img = Image.open(image_path)
             
+            # 根据概率生成炫彩结论
+            if prob > threshold_value + 0.2:
+                conclusion = "🚨 高度疑似伪造"
+                emoji = "🔴"
+            elif prob > threshold_value:
+                conclusion = "⚠️ 可能伪造"
+                emoji = "🟡"
+            else:
+                conclusion = "✅ 未检测到伪造"
+                emoji = "🟢"
+            
             return {
                 image_input: original_img,
                 output_image: processed_img,
                 original_display: original_img,
                 processed_display: processed_img,
                 fake_prob: prob,
-                result_text: f"伪造概率: {prob:.4f}",
+                result_text: f"{emoji} {conclusion} (概率: {prob:.2f})",
                 threshold: threshold_value
             }
         except Exception as e:
@@ -135,207 +530,60 @@ with gr.Blocks(title="Loupe图像伪造检测系统", theme=gr.themes.Soft()) as
                 original_display: None,
                 processed_display: None,
                 fake_prob: 0.0,
-                result_text: "处理错误",
+                result_text: f"❌ 示例处理错误: {str(e)}",
                 threshold: threshold_value
             }
-    
+
+    def clear_all():
+        """清除所有输入输出"""
+        return {
+            image_input: None,
+            output_image: None,
+            original_display: None,
+            processed_display: None,
+            fake_prob: 0.0,
+            result_text: "🧹 已清除所有数据"
+        }
+
+    def update_selected_index(evt: gr.SelectData):
+        """更新选中的图片索引"""
+        return evt.index
+
+    # 交互逻辑
     upload_button.click(
         process_image,
         [image_input, threshold],
         [output_image, original_display, processed_display, fake_prob, result_text]
     )
     
+    # 示例图片选择事件
+    example_gallery.select(
+        update_selected_index,
+        None,
+        selected_index
+    )
+    
+    # 示例图片检测按钮点击事件
     example_button.click(
         process_example,
-        [example_dropdown, threshold],
+        [example_gallery, selected_index, threshold],
         [image_input, output_image, original_display, processed_display, fake_prob, result_text, threshold]
     )
     
     save_button.click(
-        lambda img: img.save("result.jpg") if img else None,
+        lambda img: (img.save("result.jpg"), "💾 结果已保存为 result.jpg")[1] if img else "❌ 没有图像可保存",
         [output_image],
-        None
+        None,
+        api_name="save_result"
+    )
+    
+    clear_button.click(
+        clear_all,
+        [],
+        [image_input, output_image, original_display, processed_display, fake_prob, result_text]
     )
 
 if __name__ == "__main__":
-    demo.launch(server_name="0.0.0.0", server_port=7860)
-
-# import gradio as gr
-# import os
-# import tempfile
-# import numpy as np
-# from PIL import Image
-# from src.predict import process_single_image
-# import sys
-# sys.path.insert(0, "./src")  # 确保src目录在路径中
-# # 可以处理无mask的图像 也可以处理有mask的两张图像
-
-
-# # 获取图像文件夹中的图片列表
-# def get_example_images(folder_path="/home/xxw/Loupe/ffhq"):
-#     return [os.path.join(folder_path, f) for f in os.listdir(folder_path) 
-#             if f.lower().endswith(('.png', '.jpg', '.jpeg'))]
-
-
-# # 创建Gradio界面
-# with gr.Blocks(title="Loupe图像伪造检测系统", theme=gr.themes.Soft()) as demo:
-#     # 标题和描述
-#     gr.Markdown("""
-#     # Loupe🕵️‍♂️ 图像伪造检测系统
-#     ### 上传图像或从示例中选择，系统将检测图像中的伪造区域
-#     """)
-    
-#     with gr.Row():
-#         # 左侧面板 - 原始图像
-#         with gr.Column(scale=1):
-#             with gr.Tab("上传图像"):
-#                 image_input = gr.Image(type="pil", label="原始图像")
-#                 upload_button = gr.Button("检测伪造", variant="primary")
-            
-#             with gr.Tab("选择示例"):
-#                 example_images = get_example_images()
-#                 example_dropdown = gr.Dropdown(
-#                     choices=example_images,
-#                     label="选择示例图像",
-#                     value=example_images[0] if example_images else None
-#                 )
-#                 example_button = gr.Button("检测示例", variant="secondary")
-            
-#             with gr.Accordion("高级选项", open=False):
-#                 threshold = gr.Slider(0, 1, value=0.5, 
-#                                     label="检测阈值", 
-#                                     info="调整伪造检测的敏感度")
-#                 processing_mode = gr.Radio(
-#                     ["快速模式", "精确模式"],
-#                     value="快速模式",
-#                     label="处理模式"
-#                 )
-        
-#         # 右侧输出面板
-#         with gr.Column(scale=1):
-#             gr.Markdown("### 检测结果")
-#             with gr.Tabs():
-#                 with gr.Tab("处理后的图像"):
-#                     output_image = gr.Image(label="伪造检测结果", interactive=False)
-                
-#                 with gr.Tab("对比视图"):
-#                     with gr.Row():
-#                         original_display = gr.Image(label="原始图像", interactive=False)
-#                         processed_display = gr.Image(label="处理后图像", interactive=False)
-            
-#             with gr.Group():
-#                 with gr.Row():
-#                     fake_prob = gr.Number(label="伪造概率", precision=2)
-#                     result_label = gr.Label(label="检测结论")
-            
-#             save_button = gr.Button("保存结果", variant="secondary")
-    
-#     # 底部信息
-#     gr.Markdown("""
-#     ---
-#     ### 关于
-#     - **技术**: Forgery Image Detection and Localization.
-#     - **版本**: 1.0.0
-#     - **开发者**: xxw/teleai EVOL lab
-#     """)
-
-#     # 定义处理函数
-#     def process_image(image, threshold_value):
-#         # 创建一个临时文件保存上传的图像
-#         with tempfile.NamedTemporaryFile(suffix=".png", delete=False) as tmp_file:
-#             image_path = tmp_file.name
-#             image.save(image_path)
-
-#         try:
-#             # 调用你的处理函数
-#             processed_img, cls_probs = process_single_image(image_path)
-
-#             # 获取伪造概率（假设cls_probs是一个数组，取第一个值）
-#             prob = float(cls_probs[0]) if cls_probs else 0.0
-
-#             # 确定结果 - 修改为返回字典格式
-#             result = {
-#                 "label": "伪造图像" if prob > threshold_value else "真实图像",
-#                 "confidences": [
-#                     {"label": "伪造", "confidence": prob},
-#                     {"label": "真实", "confidence": 1 - prob}
-#                 ]
-#             }
-
-#             return {
-#                 output_image: processed_img,
-#                 original_display: image,
-#                 processed_display: processed_img,
-#                 fake_prob: prob,
-#                 result_label: result  # 使用正确的字典格式
-#             }
-#         finally:
-#             # 清理临时文件
-#             if os.path.exists(image_path):
-#                 os.unlink(image_path)
-    
-#     def process_example(image_path, threshold_value):
-#         # 直接调用你的处理函数
-#         processed_img, cls_probs = process_single_image(image_path)
-
-#         # 获取伪造概率
-#         prob = float(cls_probs[0]) if cls_probs else 0.0
-
-#         # 确定结果 - 修改为返回字典格式
-#         result = {
-#             "label": "伪造图像" if prob > threshold_value else "真实图像",
-#             "confidences": [
-#                 {"label": "伪造", "confidence": prob},
-#                 {"label": "真实", "confidence": 1 - prob}
-#             ]
-#         }
-
-#         # 打开原始图像用于显示
-#         original_img = Image.open(image_path)
-
-#         return {
-#             image_input: original_img,
-#             output_image: processed_img,
-#             original_display: original_img,
-#             processed_display: processed_img,
-#             fake_prob: prob,
-#             result_label: result,  # 使用正确的字典格式
-#             threshold: threshold_value
-#         }
-    
-#     # 修改绑定事件
-#     upload_button.click(
-#         fn=process_image,
-#         inputs=[image_input, threshold],
-#         outputs=[output_image, original_display, processed_display, fake_prob, result_label]
-#     )
-    
-#     def load_example_image(example_path):
-#         try:
-#             return Image.open(example_path)
-#         except:
-#             return None
-    
-#     example_button.click(
-#         fn=process_example,
-#         inputs=[example_dropdown, threshold],
-#         outputs=[image_input, output_image, original_display, processed_display, fake_prob, result_label, threshold]
-#     )
-    
-#     save_button.click(
-#         fn=lambda img: (img.save("result.jpg") if img else None) or "结果已保存!",
-#         inputs=[output_image],
-#         outputs=gr.Textbox(visible=True, label="保存状态"),
-#         api_name="save_result"
-#     )
-
-
-# # def greet(name):
-# #     return "Hello " + name + "!!"
-
-# # demo = gr.Interface(fn=greet, inputs="text", outputs="text")
-# # demo.launch()
-
-# # 启动应用
-# if __name__ == "__main__":
-#     demo.launch(server_name="0.0.0.0", server_port=7860)
+    demo.launch(
+        favicon_path="./favicon.ico" if os.path.exists("./favicon.ico") else None
+    )