Upload 3 files

app.py

@@ -0,0 +1,632 @@
+"""
+目标检测 Web 应用 - Gradio 界面
+基于 YOLOv8 预训练模型 (COCO 80类)
+"""
+
+import os
+import sys
+import gradio as gr
+import cv2
+
+sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
+
+from detector import Detector
+from visualize import draw_detections as draw_boxes
+
+# ============================================================
+# 全局配置
+# ============================================================
+MODEL_NAME = 'yolov8s.pt'
+
+print("=" * 60)
+print("智能目标检测系统启动中...")
+print(f"模型: YOLOv8 Small (COCO 80类)")
+print("=" * 60)
+
+detector = Detector(model_name=MODEL_NAME)
+
+
+# ============================================================
+# 样式定义
+# ============================================================
+CUSTOM_CSS = """
+/* === 全局样式 === */
+@import url('https://fonts.googleapis.com/css2?family=Inter:wght@300;400;500;600;700;800&display=swap');
+
+* {
+    font-family: 'Inter', 'Segoe UI', 'PingFang SC', 'Microsoft YaHei', sans-serif !important;
+}
+
+.gradio-container {
+    max-width: 1400px !important;
+    margin: 0 auto !important;
+    background: linear-gradient(135deg, #f5f7fa 0%, #e4e9f0 100%) !important;
+}
+
+body {
+    background: linear-gradient(135deg, #0f172a 0%, #1e293b 50%, #0f172a 100%) !important;
+}
+
+/* === 主容器 === */
+.main-container {
+    background: rgba(255, 255, 255, 0.95);
+    backdrop-filter: blur(20px);
+    border-radius: 24px;
+    box-shadow: 0 8px 32px rgba(0, 0, 0, 0.08), 0 2px 8px rgba(0, 0, 0, 0.04);
+    padding: 40px;
+    margin: 20px;
+    border: 1px solid rgba(255, 255, 255, 0.6);
+}
+
+/* === 标题区域 === */
+.header-section {
+    text-align: center;
+    margin-bottom: 36px;
+    position: relative;
+}
+
+.header-badge {
+    display: inline-block;
+    background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
+    color: white;
+    padding: 6px 20px;
+    border-radius: 50px;
+    font-size: 13px;
+    font-weight: 600;
+    letter-spacing: 1.5px;
+    text-transform: uppercase;
+    margin-bottom: 16px;
+}
+
+.header-title {
+    font-size: 42px;
+    font-weight: 800;
+    background: linear-gradient(135deg, #1a1a2e 0%, #16213e 50%, #0f3460 100%);
+    -webkit-background-clip: text;
+    -webkit-text-fill-color: transparent;
+    background-clip: text;
+    margin: 12px 0;
+    letter-spacing: -0.5px;
+}
+
+.header-subtitle {
+    font-size: 16px;
+    color: #64748b;
+    font-weight: 400;
+    max-width: 600px;
+    margin: 0 auto;
+    line-height: 1.6;
+}
+
+/* === 状态指示器 === */
+.status-bar {
+    display: flex;
+    align-items: center;
+    justify-content: center;
+    gap: 24px;
+    margin-bottom: 32px;
+    flex-wrap: wrap;
+}
+
+.status-item {
+    display: flex;
+    align-items: center;
+    gap: 8px;
+    background: white;
+    padding: 10px 20px;
+    border-radius: 16px;
+    box-shadow: 0 2px 12px rgba(0, 0, 0, 0.04);
+    font-size: 14px;
+    font-weight: 500;
+    color: #334155;
+}
+
+.status-dot {
+    width: 10px;
+    height: 10px;
+    border-radius: 50%;
+    background: #22c55e;
+    box-shadow: 0 0 8px rgba(34, 197, 94, 0.4);
+    animation: pulse-dot 2s infinite;
+}
+
+@keyframes pulse-dot {
+    0%, 100% { opacity: 1; transform: scale(1); }
+    50% { opacity: 0.6; transform: scale(0.85); }
+}
+
+.status-dot.warn {
+    background: #f59e0b;
+    box-shadow: 0 0 8px rgba(245, 158, 11, 0.4);
+}
+
+/* === 卡片容器 === */
+.card {
+    background: white;
+    border-radius: 20px;
+    padding: 28px;
+    box-shadow: 0 4px 20px rgba(0, 0, 0, 0.04);
+    border: 1px solid rgba(0, 0, 0, 0.06);
+    transition: all 0.3s ease;
+}
+
+.card:hover {
+    box-shadow: 0 8px 30px rgba(0, 0, 0, 0.08);
+    transform: translateY(-1px);
+}
+
+.card-title {
+    font-size: 18px;
+    font-weight: 700;
+    color: #1e293b;
+    margin-bottom: 20px;
+    display: flex;
+    align-items: center;
+    gap: 10px;
+}
+
+.card-title .icon {
+    width: 36px;
+    height: 36px;
+    border-radius: 12px;
+    display: flex;
+    align-items: center;
+    justify-content: center;
+    font-size: 18px;
+}
+
+.icon-upload { background: linear-gradient(135deg, #dbeafe, #bfdbfe); }
+.icon-sliders { background: linear-gradient(135deg, #fef3c7, #fde68a); }
+.icon-result { background: linear-gradient(135deg, #dcfce7, #bbf7d0); }
+.icon-stats { background: linear-gradient(135deg, #fce7f3, #fbcfe8); }
+
+/* === 按钮样式 === */
+.detect-btn {
+    background: linear-gradient(135deg, #667eea 0%, #764ba2 100%) !important;
+    border: none !important;
+    color: white !important;
+    font-weight: 700 !important;
+    font-size: 16px !important;
+    padding: 14px 32px !important;
+    border-radius: 14px !important;
+    box-shadow: 0 4px 16px rgba(102, 126, 234, 0.4) !important;
+    transition: all 0.3s ease !important;
+    letter-spacing: 0.5px !important;
+    width: 100% !important;
+}
+
+.detect-btn:hover {
+    transform: translateY(-2px) !important;
+    box-shadow: 0 8px 25px rgba(102, 126, 234, 0.5) !important;
+    background: linear-gradient(135deg, #5a6fd6 0%, #6a3f9e 100%) !important;
+}
+
+.detect-btn:active {
+    transform: translateY(0) !important;
+}
+
+/* === 滑块样式 === */
+input[type="range"] {
+    accent-color: #667eea !important;
+    height: 6px !important;
+}
+
+/* === 图片区域 === */
+.image-container {
+    border-radius: 16px;
+    overflow: hidden;
+    border: 2px dashed #e2e8f0;
+    transition: all 0.3s ease;
+    background: #f8fafc;
+}
+
+.image-container:has(img) {
+    border: 2px solid #e2e8f0;
+}
+
+.image-container:hover {
+    border-color: #667eea;
+}
+
+/* === 检测摘要卡片 === */
+.detection-summary {
+    background: linear-gradient(135deg, #f8fafc 0%, #f1f5f9 100%);
+    border-radius: 16px;
+    padding: 20px;
+    border: 1px solid #e2e8f0;
+    min-height: 80px;
+}
+
+.detection-item {
+    display: inline-flex;
+    align-items: center;
+    gap: 6px;
+    background: white;
+    padding: 6px 14px;
+    border-radius: 50px;
+    margin: 4px;
+    font-size: 13px;
+    font-weight: 500;
+    box-shadow: 0 2px 8px rgba(0, 0, 0, 0.04);
+    transition: all 0.2s ease;
+}
+
+.detection-item:hover {
+    transform: translateY(-1px);
+    box-shadow: 0 4px 12px rgba(0, 0, 0, 0.08);
+}
+
+.detection-count {
+    background: linear-gradient(135deg, #667eea, #764ba2);
+    color: white;
+    padding: 2px 10px;
+    border-radius: 50px;
+    font-weight: 700;
+    font-size: 12px;
+}
+
+.no-detection {
+    text-align: center;
+    color: #94a3b8;
+    font-size: 15px;
+    padding: 24px;
+}
+
+.no-detection .icon-large {
+    font-size: 48px;
+    display: block;
+    margin-bottom: 12px;
+    opacity: 0.5;
+}
+
+/* === 统计面板 === */
+.stats-grid {
+    display: grid;
+    grid-template-columns: repeat(auto-fit, minmax(100px, 1fr));
+    gap: 12px;
+}
+
+.stat-box {
+    background: linear-gradient(135deg, #f8fafc, #f1f5f9);
+    border-radius: 14px;
+    padding: 16px;
+    text-align: center;
+    border: 1px solid #e2e8f0;
+    transition: all 0.2s ease;
+}
+
+.stat-box:hover {
+    background: linear-gradient(135deg, #eef2ff, #e0e7ff);
+    border-color: #c7d2fe;
+}
+
+.stat-number {
+    font-size: 32px;
+    font-weight: 800;
+    background: linear-gradient(135deg, #667eea, #764ba2);
+    -webkit-background-clip: text;
+    -webkit-text-fill-color: transparent;
+    background-clip: text;
+}
+
+.stat-label {
+    font-size: 12px;
+    color: #64748b;
+    font-weight: 500;
+    margin-top: 4px;
+    text-transform: uppercase;
+    letter-spacing: 0.5px;
+}
+
+/* === 置信度颜色 === */
+.conf-high { color: #22c55e; font-weight: 700; }
+.conf-medium { color: #f59e0b; font-weight: 700; }
+.conf-low { color: #ef4444; font-weight: 700; }
+
+/* === 页脚 === */
+.footer {
+    text-align: center;
+    margin-top: 48px;
+    padding: 28px;
+    color: #94a3b8;
+    font-size: 13px;
+    border-top: 1px solid #e2e8f0;
+}
+
+.footer-links {
+    display: flex;
+    justify-content: center;
+    gap: 20px;
+    margin-bottom: 12px;
+}
+
+.footer-link {
+    color: #667eea;
+    text-decoration: none;
+    font-weight: 500;
+    transition: color 0.2s;
+}
+
+.footer-link:hover {
+    color: #764ba2;
+}
+
+/* === 响应式 === */
+@media (max-width: 768px) {
+    .header-title { font-size: 28px; }
+    .stats-grid { grid-template-columns: repeat(2, 1fr); }
+}
+"""
+
+# ============================================================
+# HTML 组件
+# ============================================================
+HEADER_HTML = """
+<div class="header-section">
+    <div class="header-badge">🔬 Deep Learning · Object Detection</div>
+    <h1 class="header-title">智能目标检测系统</h1>
+    <p class="header-subtitle">
+        基于 <strong>YOLOv8</strong> 深度学习模型，支持 80 类日常物体实时检测
+        —— 上传图片，即刻识别
+    </p>
+    <div class="status-bar">
+        <div class="status-item">
+            <span class="status-dot"></span>
+            <span>YOLOv8 Small · 已就绪</span>
+        </div>
+        <div class="status-item">
+            <span>🎯</span>
+            <span>80 类目标</span>
+        </div>
+        <div class="status-item">
+            <span>⚡</span>
+            <span>实时推理</span>
+        </div>
+    </div>
+</div>
+"""
+
+FOOTER_HTML = """
+<div class="footer">
+    <div class="footer-links">
+        <a class="footer-link" href="https://docs.ultralytics.com/" target="_blank">📚 YOLOv8 文档</a>
+        <a class="footer-link" href="https://cocodataset.org/" target="_blank">📊 COCO 数据集</a>
+        <a class="footer-link" href="https://github.com/ultralytics/ultralytics" target="_blank">💻 GitHub</a>
+    </div>
+    <p>© 2026 深度学习期末综合设计 · Powered by PyTorch & Gradio</p>
+</div>
+"""
+
+
+# ============================================================
+# 辅助函数
+# ============================================================
+def get_confidence_class(score):
+    """根据置信度返回 CSS 类名"""
+    if score >= 0.7:
+        return "conf-high"
+    elif score >= 0.4:
+        return "conf-medium"
+    else:
+        return "conf-low"
+
+
+def build_summary_html(detections, class_names_en, class_names_zh):
+    """构建检测摘要 HTML"""
+    if not detections:
+        return """
+        <div class="no-detection">
+            <span class="icon-large">🔍</span>
+            <p>未检测到任何目标</p>
+            <p style="font-size:13px;color:#94a3b8;">尝试降低置信度阈值或上传其他图片</p>
+        </div>
+        """
+
+    # 统计信息
+    class_counts = {}
+    for det in detections:
+        cls_id = int(det[5])
+        en_name = class_names_en.get(cls_id, f"cls_{cls_id}")
+        zh_name = class_names_zh.get(cls_id, f"未知")
+        key = (en_name, zh_name)
+        class_counts[key] = class_counts.get(key, 0) + 1
+
+    total = len(detections)
+    unique_classes = len(class_counts)
+
+    # 构建 HTML
+    html = '<div style="margin-bottom:20px;">'
+
+    # 统计面板
+    html += '<div class="stats-grid" style="margin-bottom:20px;">'
+    html += f'<div class="stat-box"><div class="stat-number">{total}</div><div class="stat-label">检测目标</div></div>'
+    html += f'<div class="stat-box"><div class="stat-number">{unique_classes}</div><div class="stat-label">类别数</div></div>'
+    top_score = max(d[4] for d in detections) if detections else 0
+    html += f'<div class="stat-box"><div class="stat-number">{top_score:.0%}</div><div class="stat-label">最高置信度</div></div>'
+    avg_score = sum(d[4] for d in detections) / total if total > 0 else 0
+    html += f'<div class="stat-box"><div class="stat-number">{avg_score:.0%}</div><div class="stat-label">平均置信度</div></div>'
+    html += '</div>'
+
+    # 检测列表
+    html += '<div style="font-size:14px;font-weight:600;color:#475569;margin-bottom:10px;">📋 检测详情</div>'
+    html += '<div style="display:flex;flex-wrap:wrap;gap:6px;">'
+    for (en_name, zh_name), count in sorted(class_counts.items(), key=lambda x: -x[1]):
+        html += (
+            f'<div class="detection-item">'
+            f'<span>{zh_name}</span>'
+            f'<span style="color:#94a3b8;font-size:11px;">{en_name}</span>'
+            f'<span class="detection-count">×{count}</span>'
+            f'</div>'
+        )
+    html += '</div>'
+
+    # 置信度分布
+    html += '<div style="margin-top:16px;font-size:12px;color:#94a3b8;">'
+    high_conf = sum(1 for d in detections if d[4] >= 0.7)
+    mid_conf = sum(1 for d in detections if 0.4 <= d[4] < 0.7)
+    low_conf = sum(1 for d in detections if d[4] < 0.4)
+    html += f'🟢 高置信度 (≥70%): {high_conf} 个 · '
+    html += f'🟡 中置信度 (40-70%): {mid_conf} 个 · '
+    html += f'🔴 低置信度 (&lt;40%): {low_conf} 个'
+    html += '</div>'
+
+    html += '</div>'
+
+    return html
+
+
+def build_empty_summary_html():
+    """空检测的 HTML 占位"""
+    return """
+    <div class="no-detection">
+        <span class="icon-large">📸</span>
+        <p>等待上传图片...</p>
+        <p style="font-size:13px;color:#94a3b8;">上传图片后自动检测，或点击"开始检测"</p>
+    </div>
+    """
+
+
+# ============================================================
+# 检测核心函数
+# ============================================================
+def detect_and_render(image, conf_thres=0.25, iou_thres=0.45):
+    """检测 + 渲染可视化结果"""
+    if image is None:
+        return None, build_empty_summary_html()
+
+    # BGR 转换
+    image_bgr = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
+
+    # 检测
+    detections = detector.detect(image_bgr, conf_thres=conf_thres, iou_thres=iou_thres)
+
+    # 绘制边界框
+    annotated = draw_boxes(image, detections, detector.class_names, detector.colors)
+
+    # 构建 HTML 摘要
+    summary_html = build_summary_html(
+        detections,
+        detector.class_names_en,
+        detector.class_names_zh
+    )
+
+    return annotated, summary_html
+
+
+# ============================================================
+# 构建界面
+# ============================================================
+def create_ui():
+    with gr.Blocks(title="智能目标检测系统 · YOLOv8") as demo:
+
+        # 顶部 Header
+        gr.HTML(HEADER_HTML)
+
+        # 主体布局
+        with gr.Row(equal_height=True):
+            # === 左侧：上传区域 ===
+            with gr.Column(scale=5, min_width=320):
+                with gr.Group(elem_classes="card"):
+                    gr.HTML('<div class="card-title"><span class="icon icon-upload">📤</span> 上传图片</div>')
+
+                    input_image = gr.Image(
+                        label=None,
+                        type="numpy",
+                        sources=["upload", "clipboard", "webcam"],
+                        height=420,
+                        elem_classes="image-container",
+                        show_label=False,
+                    )
+
+                    with gr.Row():
+                        conf_slider = gr.Slider(
+                            minimum=0.05,
+                            maximum=0.95,
+                            value=0.25,
+                            step=0.05,
+                            label="🎯 置信度阈值",
+                            info="值越低检测越多（可能误检），越高越精准",
+                        )
+                        iou_slider = gr.Slider(
+                            minimum=0.1,
+                            maximum=0.9,
+                            value=0.45,
+                            step=0.05,
+                            label="📐 IoU 阈值",
+                            info="重叠框过滤强度，越低去重越强",
+                        )
+
+                    detect_btn = gr.Button(
+                        "🚀 开始检测",
+                        variant="primary",
+                        size="lg",
+                        elem_classes="detect-btn",
+                    )
+
+            # === 右侧：结果区域 ===
+            with gr.Column(scale=5, min_width=320):
+                with gr.Group(elem_classes="card"):
+                    gr.HTML('<div class="card-title"><span class="icon icon-result">📊</span> 检测结果</div>')
+                    output_image = gr.Image(
+                        label=None,
+                        type="numpy",
+                        height=420,
+                        elem_classes="image-container",
+                        show_label=False,
+                    )
+
+                with gr.Group(elem_classes="card"):
+                    gr.HTML('<div class="card-title"><span class="icon icon-stats">📋</span> 检测摘要</div>')
+                    output_summary = gr.HTML(
+                        value=build_empty_summary_html(),
+                        elem_classes="detection-summary",
+                    )
+
+        # 示例图片
+        with gr.Accordion("📸 示例图片（点击加载测试）", open=False):
+            gr.Examples(
+                examples=[
+                    ["https://ultralytics.com/images/bus.jpg", 0.25, 0.45],
+                    ["https://ultralytics.com/images/zidane.jpg", 0.25, 0.45],
+                ],
+                inputs=[input_image, conf_slider, iou_slider],
+                label=None,
+            )
+
+        # 页脚
+        gr.HTML(FOOTER_HTML)
+
+        # === 事件绑定 ===
+        detect_btn.click(
+            fn=detect_and_render,
+            inputs=[input_image, conf_slider, iou_slider],
+            outputs=[output_image, output_summary],
+        )
+
+        input_image.change(
+            fn=detect_and_render,
+            inputs=[input_image, conf_slider, iou_slider],
+            outputs=[output_image, output_summary],
+        )
+
+    return demo
+
+
+# ============================================================
+# 启动
+# ============================================================
+if __name__ == '__main__':
+    demo = create_ui()
+    demo.queue(max_size=20, default_concurrency_limit=5)
+    demo.launch(
+        server_name="0.0.0.0",
+        server_port=7860,
+        share=False,
+        show_error=True,
+        css=CUSTOM_CSS,
+        head="""
+        <meta name="viewport" content="width=device-width, initial-scale=1.0">
+        <meta name="description" content="基于 YOLOv8 的智能目标检测系统">
+        """,
+        favicon_path=None,
+    )

detector.py

@@ -0,0 +1,164 @@
+"""
+目标检测器模块
+基于 ultralytics YOLOv8 预训练模型 (COCO 80类)
+"""
+
+import numpy as np
+import cv2
+from ultralytics import YOLO
+import os
+
+
+# COCO 数据集 80 类中文名称映射
+COCO_CLASSES_ZH = {
+    0: '人', 1: '自行车', 2: '汽车', 3: '摩托车', 4: '飞机',
+    5: '公交车', 6: '火车', 7: '卡车', 8: '船', 9: '红绿灯',
+    10: '消火栓', 11: '停车标志', 12: '停车计时器', 13: '长椅', 14: '鸟',
+    15: '猫', 16: '狗', 17: '马', 18: '羊', 19: '牛',
+    20: '大象', 21: '熊', 22: '斑马', 23: '长颈鹿', 24: '背包',
+    25: '雨伞', 26: '手提包', 27: '领带', 28: '行李箱', 29: '飞盘',
+    30: '滑雪板', 31: '雪板', 32: '球', 33: '风筝', 34: '棒球棒',
+    35: '手套', 36: '滑板', 37: '冲浪板', 38: '网球拍', 39: '瓶子',
+    40: '酒杯', 41: '杯子', 42: '叉子', 43: '刀', 44: '勺子',
+    45: '碗', 46: '香蕉', 47: '苹果', 48: '三明治', 49: '橙子',
+    50: '西兰花', 51: '胡萝卜', 52: '热狗', 53: '披萨', 54: '甜甜圈',
+    55: '蛋糕', 56: '椅子', 57: '沙发', 58: '盆栽', 59: '床',
+    60: '餐桌', 61: '马桶', 62: '显示器', 63: '笔记本', 64: '鼠标',
+    65: '遥控器', 66: '键盘', 67: '手机', 68: '微波炉', 69: '烤箱',
+    70: '烤面包机', 71: '水槽', 72: '冰箱', 73: '书', 74: '钟',
+    75: '花瓶', 76: '剪刀', 77: '泰迪熊', 78: '吹风机', 79: '牙刷'
+}
+
+COCO_CLASSES_EN = {
+    0: 'person', 1: 'bicycle', 2: 'car', 3: 'motorcycle', 4: 'airplane',
+    5: 'bus', 6: 'train', 7: 'truck', 8: 'boat', 9: 'traffic light',
+    10: 'fire hydrant', 11: 'stop sign', 12: 'parking meter', 13: 'bench', 14: 'bird',
+    15: 'cat', 16: 'dog', 17: 'horse', 18: 'sheep', 19: 'cow',
+    20: 'elephant', 21: 'bear', 22: 'zebra', 23: 'giraffe', 24: 'backpack',
+    25: 'umbrella', 26: 'handbag', 27: 'tie', 28: 'suitcase', 29: 'frisbee',
+    30: 'skis', 31: 'snowboard', 32: 'sports ball', 33: 'kite', 34: 'baseball bat',
+    35: 'baseball glove', 36: 'skateboard', 37: 'surfboard', 38: 'tennis racket', 39: 'bottle',
+    40: 'wine glass', 41: 'cup', 42: 'fork', 43: 'knife', 44: 'spoon',
+    45: 'bowl', 46: 'banana', 47: 'apple', 48: 'sandwich', 49: 'orange',
+    50: 'broccoli', 51: 'carrot', 52: 'hot dog', 53: 'pizza', 54: 'donut',
+    55: 'cake', 56: 'chair', 57: 'couch', 58: 'potted plant', 59: 'bed',
+    60: 'dining table', 61: 'toilet', 62: 'tv', 63: 'laptop', 64: 'mouse',
+    65: 'remote', 66: 'keyboard', 67: 'cell phone', 68: 'microwave', 69: 'oven',
+    70: 'toaster', 71: 'sink', 72: 'refrigerator', 73: 'book', 74: 'clock',
+    75: 'vase', 76: 'scissors', 77: 'teddy bear', 78: 'hair drier', 79: 'toothbrush'
+}
+
+
+class Detector:
+    """YOLOv8 目标检测器（使用 ultralytics 预训练模型）"""
+
+    def __init__(self, model_name='yolov8n.pt', device=None):
+        """
+        初始化检测器
+
+        Args:
+            model_name: 模型名称或路径
+                - 'yolov8n.pt'  Nano (最快, ~6MB)
+                - 'yolov8s.pt'  Small (~22MB)
+                - 'yolov8m.pt'  Medium (~52MB)
+            device: 推理设备 ('cuda', 'cpu', 或 None 自动选择)
+        """
+        self.device = device or ('cuda' if cv2.cuda.getCudaEnabledDeviceCount() > 0 else 'cpu')
+        print(f"[INFO] 使用设备: {self.device}")
+
+        # 加载模型
+        # 优先使用本地模型文件，否则尝试从镜像下载
+        local_model = os.path.join(os.path.dirname(os.path.dirname(__file__)), 'models', model_name)
+        if os.path.exists(local_model):
+            model_name = local_model
+        elif os.path.exists(model_name):
+            pass  # 使用用户指定的路径
+        else:
+            # 尝试从 Hugging Face 镜像下载
+            print(f"[INFO] 本地模型不存在，尝试从镜像下载: {model_name}")
+            try:
+                os.environ.setdefault('HF_ENDPOINT', 'https://hf-mirror.com')
+                from huggingface_hub import hf_hub_download
+                model_dir = os.path.join(os.path.dirname(os.path.dirname(__file__)), 'models')
+                os.makedirs(model_dir, exist_ok=True)
+                downloaded = hf_hub_download(
+                    repo_id='Ultralytics/YOLOv8',
+                    filename=model_name,
+                    local_dir=model_dir,
+                )
+                model_name = downloaded
+                print(f"[INFO] 模型下载成功: {model_name}")
+            except Exception as e:
+                print(f"[WARN] 镜像下载失败: {e}，将尝试 ultralytics 默认下载")
+
+        self.model_path = model_name
+        self.model = YOLO(model_name)
+
+        # 类别信息
+        self.class_names_en = COCO_CLASSES_EN
+        self.class_names_zh = COCO_CLASSES_ZH
+        self.class_names = [f"{COCO_CLASSES_EN[i]} ({COCO_CLASSES_ZH[i]})" for i in range(80)]
+        self.num_classes = 80
+        self.class_names_en_only = [COCO_CLASSES_EN[i] for i in range(80)]
+
+        # 为每个类别生��颜色
+        np.random.seed(42)
+        self.colors = np.random.randint(50, 255, size=(self.num_classes, 3), dtype=np.uint8).tolist()
+
+        print(f"[INFO] 模型加载完成: {model_name}")
+        print(f"[INFO] 类别数量: {self.num_classes} (COCO)")
+
+    def detect(self, image, conf_thres=0.25, iou_thres=0.45, max_det=100):
+        """
+        对图片进行目标检测
+
+        Args:
+            image: numpy array (H, W, 3) BGR 图片
+            conf_thres: 置信度阈值
+            iou_thres: NMS IoU 阈值
+            max_det: 最大检测数量
+
+        Returns:
+            detections: list of [x1, y1, x2, y2, score, class_id]
+        """
+        # 转换为 RGB（ultralytics 需要 RGB）
+        image_rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
+
+        # 运行检测
+        results = self.model(
+            image_rgb,
+            conf=conf_thres,
+            iou=iou_thres,
+            max_det=max_det,
+            verbose=False,
+        )
+
+        # 解析结果
+        detections = []
+        if results and len(results) > 0:
+            result = results[0]
+            if result.boxes is not None and len(result.boxes) > 0:
+                boxes = result.boxes.xyxy.cpu().numpy()       # [N, 4] x1,y1,x2,y2
+                scores = result.boxes.conf.cpu().numpy()      # [N]
+                class_ids = result.boxes.cls.cpu().numpy().astype(int)  # [N]
+
+                for box, score, cls_id in zip(boxes, scores, class_ids):
+                    x1, y1, x2, y2 = box.tolist()
+                    detections.append([x1, y1, x2, y2, float(score), int(cls_id)])
+
+                # 按置信度降序
+                detections.sort(key=lambda x: x[4], reverse=True)
+
+        return detections
+
+    def get_class_name(self, class_id, lang='zh'):
+        """获取类别名称"""
+        if lang == 'zh':
+            return self.class_names_zh.get(class_id, f"未知_{class_id}")
+        return self.class_names_en.get(class_id, f"unknown_{class_id}")
+
+    def get_color(self, class_id):
+        """获取类别对应颜色"""
+        if 0 <= class_id < len(self.colors):
+            return tuple(self.colors[class_id])
+        return (255, 255, 255)

visualize.py

@@ -0,0 +1,293 @@
+"""
+检测结果可视化模块
+在图片上绘制检测框、类别标签和置信度
+"""
+
+import cv2
+import numpy as np
+from PIL import Image, ImageDraw, ImageFont
+import matplotlib.pyplot as plt
+from io import BytesIO
+
+
+def draw_detections_cv2(image, detections, class_names, colors):
+    """
+    使用 OpenCV 绘制检测结果（适合中文环境用 PIL 方案更好）
+
+    Args:
+        image: numpy array (H, W, 3) BGR
+        detections: list of [x1, y1, x2, y2, score, class_id]
+        class_names: list of class name strings
+        colors: list of (R, G, B) tuples
+
+    Returns:
+        image with boxes drawn (BGR)
+    """
+    result = image.copy()
+
+    for det in detections:
+        x1, y1, x2, y2, score, cls_id = det
+        x1, y1, x2, y2 = int(x1), int(y1), int(x2), int(y2)
+        cls_id = int(cls_id)
+
+        color = tuple(colors[cls_id]) if cls_id < len(colors) else (0, 255, 0)
+        class_name = class_names[cls_id] if cls_id < len(class_names) else f"cls_{cls_id}"
+        label = f"{class_name}: {score:.2f}"
+
+        # 绘制边界框
+        cv2.rectangle(result, (x1, y1), (x2, y2), color, 2)
+
+        # 绘制标签背景
+        (label_w, label_h), baseline = cv2.getTextSize(label, cv2.FONT_HERSHEY_SIMPLEX, 0.5, 2)
+        cv2.rectangle(result, (x1, y1 - label_h - 10), (x1 + label_w, y1), color, -1)
+
+        # 绘制标签文字
+        cv2.putText(result, label, (x1, y1 - 5), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 255, 255), 2)
+
+    return result
+
+
+def draw_detections_pil(image, detections, class_names, colors, font_path=None):
+    """
+    使用 PIL 绘制检测结果（支持中文，效果更好）
+
+    Args:
+        image: numpy array (H, W, 3) RGB
+        detections: list of [x1, y1, x2, y2, score, class_id]
+        class_names: list of class name strings
+        colors: list of (R, G, B) tuples
+        font_path: 中文字体路径（可选）
+
+    Returns:
+        image with boxes drawn (RGB, numpy array)
+    """
+    result = Image.fromarray(image.astype(np.uint8))
+    draw = ImageDraw.Draw(result)
+
+    # 尝试加载字体
+    try:
+        if font_path:
+            font = ImageFont.truetype(font_path, 16)
+            small_font = ImageFont.truetype(font_path, 12)
+        else:
+            # 尝试系统默认字体
+            font = ImageFont.truetype("arial.ttf", 16)
+            small_font = ImageFont.truetype("arial.ttf", 12)
+    except Exception:
+        font = ImageFont.load_default()
+        small_font = ImageFont.load_default()
+
+    for det in detections:
+        x1, y1, x2, y2, score, cls_id = det
+        cls_id = int(cls_id)
+
+        color = tuple(colors[cls_id]) if cls_id < len(colors) else (0, 255, 0)
+        class_name = class_names[cls_id] if cls_id < len(class_names) else f"cls_{cls_id}"
+        label = f"{class_name}: {score:.2f}"
+
+        # 绘制边界框（加粗）
+        for offset in range(2):
+            draw.rectangle(
+                [x1 - offset, y1 - offset, x2 + offset, y2 + offset],
+                outline=color,
+                width=1
+            )
+
+        # 绘制标签背景和文字
+        try:
+            text_bbox = draw.textbbox((0, 0), label, font=small_font)
+            text_w = text_bbox[2] - text_bbox[0]
+            text_h = text_bbox[3] - text_bbox[1]
+        except Exception:
+            text_w, text_h = 60, 14
+
+        label_y = max(0, y1 - text_h - 6)
+        draw.rectangle([x1, label_y, x1 + text_w + 4, label_y + text_h + 4], fill=color)
+        draw.text((x1 + 2, label_y + 2), label, fill=(255, 255, 255), font=small_font)
+
+    return np.array(result)
+
+
+def draw_detections_pil_enhanced(image, detections, class_names, colors):
+    """
+    增强版 PIL 绘制，支持阴影、圆角、半透明等美化效果
+
+    Args:
+        image: numpy array (H, W, 3) RGB
+        detections: list of [x1, y1, x2, y2, score, class_id]
+        class_names: list of class name strings
+        colors: list of (R, G, B) tuples
+
+    Returns:
+        image with beautified boxes drawn (RGB, numpy array)
+    """
+    from PIL import ImageDraw
+
+    h, w = image.shape[:2]
+    result = Image.fromarray(image.astype(np.uint8))
+    overlay = Image.new('RGBA', (w, h), (0, 0, 0, 0))
+    overlay_draw = ImageDraw.Draw(overlay)
+    draw = ImageDraw.Draw(result)
+
+    # 尝试加载字体
+    try:
+        # Windows 中文字体
+        for font_path in [
+            "C:/Windows/Fonts/msyh.ttc",
+            "C:/Windows/Fonts/simhei.ttf",
+            "C:/Windows/Fonts/msyhbd.ttc",
+            "/usr/share/fonts/truetype/noto/NotoSansCJK-Regular.ttc",
+        ]:
+            try:
+                font = ImageFont.truetype(font_path, 16)
+                small_font = ImageFont.truetype(font_path, 12)
+                break
+            except Exception:
+                continue
+        else:
+            font = ImageFont.load_default()
+            small_font = ImageFont.load_default()
+    except Exception:
+        font = ImageFont.load_default()
+        small_font = ImageFont.load_default()
+
+    # 按置信度降序绘制（高置信度在上层）
+    sorted_dets = sorted(detections, key=lambda x: x[4])
+
+    for det in sorted_dets:
+        x1, y1, x2, y2, score, cls_id = det
+        x1, y1 = max(0, int(x1)), max(0, int(y1))
+        x2, y2 = min(w, int(x2)), min(h, int(y2))
+        cls_id = int(cls_id)
+        if x2 <= x1 or y2 <= y1:
+            continue
+
+        color = tuple(colors[cls_id]) if cls_id < len(colors) else (0, 255, 0)
+        class_name = class_names[cls_id] if cls_id < len(class_names) else f"cls_{cls_id}"
+
+        # 根据置信度调整线宽
+        if score >= 0.7:
+            line_width = 4
+            shadow_offset = 3
+        elif score >= 0.4:
+            line_width = 3
+            shadow_offset = 2
+        else:
+            line_width = 2
+            shadow_offset = 1
+
+        # === 绘制阴影 ===
+        shadow_color = (0, 0, 0, 40)
+        for dx in range(shadow_offset):
+            overlay_draw.rectangle(
+                [x1 + shadow_offset, y1 + shadow_offset, x2 + shadow_offset, y2 + shadow_offset],
+                outline=shadow_color,
+                width=line_width + 2,
+            )
+
+        # === 绘制半透明填充 ===
+        fill_alpha = 25
+        if score >= 0.7:
+            fill_alpha = 35
+        fill_rgba = color + (fill_alpha,)
+        overlay_draw.rectangle([x1, y1, x2, y2], fill=fill_rgba)
+
+        # === 绘制主边界框 ===
+        # 外层（深色边缘）
+        dark_color = tuple(max(0, c - 40) for c in color)
+        overlay_draw.rectangle([x1, y1, x2, y2], outline=dark_color + (220,), width=line_width + 2)
+        # 内层（亮色）
+        overlay_draw.rectangle([x1, y1, x2, y2], outline=color + (255,), width=line_width)
+
+        # === 绘制角标（四角加粗） ===
+        corner_len = min(15, (x2 - x1) // 4, (y2 - y1) // 4)
+        for cx, cy, dx, dy in [
+            (x1, y1, 1, 1), (x2, y1, -1, 1),
+            (x1, y2, 1, -1), (x2, y2, -1, -1),
+        ]:
+            overlay_draw.line(
+                [(cx, cy), (cx + corner_len * dx, cy), (cx, cy), (cx, cy + corner_len * dy)],
+                fill=color + (230,),
+                width=line_width + 1,
+            )
+
+        # === 绘制标签 ===
+        label_text = f"{class_name} {score:.0%}"
+        try:
+            text_bbox = draw.textbbox((0, 0), label_text, font=small_font)
+            text_w = text_bbox[2] - text_bbox[0]
+            text_h = text_bbox[3] - text_bbox[1]
+        except Exception:
+            text_w, text_h = len(label_text) * 7, 14
+
+        padding = 6
+        label_bg_w = text_w + padding * 2
+        label_bg_h = text_h + padding
+
+        # 标签位置（优先放框内顶部，框太小就放框外）
+        if y1 > label_bg_h + 8:
+            label_y = y1 - label_bg_h - 2
+            label_bg_y1 = label_y
+            label_bg_y2 = y1
+        else:
+            label_y = y1 + 2
+            label_bg_y1 = y1
+            label_bg_y2 = y1 + label_bg_h
+
+        label_x = x1 + 2
+
+        # 标签背景（扩展到框外）
+        overlay_draw.rectangle(
+            [label_x, label_bg_y1, label_x + label_bg_w, label_bg_y2],
+            fill=color + (200,),
+        )
+
+        # 标签文字
+        draw.text(
+            (label_x + padding, label_y + padding // 2),
+            label_text,
+            fill=(255, 255, 255),
+            font=small_font,
+        )
+
+    # 合成
+    result = Image.alpha_composite(result.convert('RGBA'), overlay)
+    return np.array(result.convert('RGB'))
+
+
+def draw_detections(image, detections, class_names, colors):
+    """
+    综合绘制检测结果
+    """
+    try:
+        return draw_detections_pil_enhanced(image, detections, class_names, colors)
+    except Exception:
+        return draw_detections_pil(image, detections, class_names, colors)
+
+
+def create_summary_text(detections, class_names):
+    """
+    生成检测结果摘要文本
+
+    Args:
+        detections: list of [x1, y1, x2, y2, score, class_id]
+        class_names: list of class name strings
+
+    Returns:
+        summary string
+    """
+    if not detections:
+        return "未检测到任何目标。"
+
+    # 统计各类别数量
+    class_counts = {}
+    for det in detections:
+        cls_id = int(det[5])
+        class_name = class_names[cls_id] if cls_id < len(class_names) else f"未知_{cls_id}"
+        class_counts[class_name] = class_counts.get(class_name, 0) + 1
+
+    lines = [f"共检测到 {len(detections)} 个目标:"]
+    for name, count in class_counts.items():
+        lines.append(f"  • {name}: {count} 个")
+
+    return "\n".join(lines)