update

app.py

@@ -0,0 +1,91 @@
+import gradio as gr
+import torch
+from PIL import Image
+import io
+import pandas as pd
+
+from model import RadarDetectionModel
+from feature_extraction import (calculate_amplitude, classify_amplitude,
+                                calculate_distribution_range, classify_distribution_range,
+                                calculate_attenuation_rate, classify_attenuation_rate,
+                                count_reflections, classify_reflections)
+from report_generation import generate_report
+from utils import plot_detection, render_report
+from database import save_report, get_report_history
+
+model = RadarDetectionModel()
+
+
+def process_image(image):
+    detection_result = model.detect(image)
+
+    np_image = np.array(image)
+    amplitude = calculate_amplitude(np_image)
+    amplitude_class = classify_amplitude(amplitude)
+
+    box = detection_result['boxes'][0].tolist()
+    distribution_range = calculate_distribution_range(box)
+    distribution_class = classify_distribution_range(distribution_range)
+
+    attenuation_rate = calculate_attenuation_rate(np_image)
+    attenuation_class = classify_attenuation_rate(attenuation_rate)
+
+    reflection_count = count_reflections(np_image)
+    reflection_class = classify_reflections(reflection_count)
+
+    features = {
+        "振幅": amplitude_class,
+        "分布范围": distribution_class,
+        "衰减速度": attenuation_class,
+        "反射次数": reflection_class
+    }
+
+    report = generate_report(detection_result, image, features)
+
+    detection_image = plot_detection(image, detection_result)
+
+    save_report(report)
+
+    return detection_image, report
+
+
+def analyze_radar_image(image):
+    detection_image, report = process_image(image)
+    report_html = render_report(report)
+    return detection_image, report_html
+
+
+def display_history():
+    reports = get_report_history()
+    history_html = "<div class='history-container'><h3>历史记录</h3>"
+    for report in reports:
+        history_html += f"""
+        <div class='history-item'>
+            <p><strong>报告ID:</strong> {report.report_id}</p>
+            <p><strong>缺陷类型:</strong> {report.defect_type}</p>
+            <p><strong>描述:</strong> {report.description}</p>
+            <p><strong>创建时间:</strong> {report.created_at}</p>
+        </div>
+        """
+    history_html += "</div>"
+    return history_html
+
+
+with gr.Blocks(css="static/style.css") as iface:
+    gr.Markdown("# 雷达图谱分析系统")
+    with gr.Row():
+        with gr.Column(scale=1):
+            input_image = gr.Image(type="pil", label="上传雷达图谱")
+            analyze_button = gr.Button("分析")
+        with gr.Column(scale=2):
+            output_image = gr.Image(type="pil", label="检测结果")
+            output_report = gr.HTML(label="分析报告")
+
+    history_button = gr.Button("查看历史记录")
+    history_output = gr.HTML()
+
+    analyze_button.click(analyze_radar_image, inputs=[
+                         input_image], outputs=[output_image, output_report])
+    history_button.click(display_history, inputs=[], outputs=[history_output])
+
+iface.launch()

config.py

@@ -0,0 +1,13 @@
+import os
+
+BASE_DIR = os.path.dirname(os.path.abspath(__file__))
+
+MODEL_NAME = "Extremely4606/paligemma_9_19"
+DATABASE_URL = f"sqlite:///{os.path.join(BASE_DIR, 'radar_reports.db')}"
+
+AMPLITUDE_THRESHOLD = 128
+DISTRIBUTION_THRESHOLD = 1000
+ATTENUATION_THRESHOLD = 0.5
+REFLECTION_THRESHOLD = 2
+
+HISTORY_LIMIT = 20

data_visualization.py

@@ -0,0 +1,54 @@
+import plotly.graph_objs as go
+import pandas as pd
+from sklearn.decomposition import PCA
+
+
+def create_feature_radar_chart(features):
+    categories = ['振幅', '分布范围', '衰减速度', '反射次数']
+    values = [features[f'{cat}值'] for cat in categories]
+
+    fig = go.Figure(data=go.Scatterpolar(
+        r=values,
+        theta=categories,
+        fill='toself'
+    ))
+
+    fig.update_layout(
+        polar=dict(
+            radialaxis=dict(visible=True, range=[0, max(values)])
+        ),
+        showlegend=False
+    )
+
+    return fig.to_html(full_html=False)
+
+
+def create_pca_visualization(reports):
+    features = ['振幅值', '分布范围值', '衰减速度值', '反射次数值']
+    X = pd.DataFrame([r.features for r in reports])[features]
+
+    pca = PCA(n_components=2)
+    pca_result = pca.fit_transform(X)
+
+    df = pd.DataFrame(data=pca_result, columns=['PC1', 'PC2'])
+    df['缺陷类型'] = [r.defect_type for r in reports]
+
+    fig = go.Figure(data=go.Scatter(
+        x=df['PC1'],
+        y=df['PC2'],
+        mode='markers',
+        marker=dict(
+            size=10,
+            color=df['缺陷类型'].map({'空洞': 'red', '裂缝': 'blue'}),
+            opacity=0.8
+        ),
+        text=df['缺陷类型']
+    ))
+
+    fig.update_layout(
+        title='PCA of Radar Features',
+        xaxis_title='Principal Component 1',
+        yaxis_title='Principal Component 2'
+    )
+
+    return fig.to_html(full_html=False)

database.py

@@ -0,0 +1,44 @@
+from sqlalchemy import create_engine, Column, Integer, String, DateTime, Float, JSON
+from sqlalchemy.ext.declarative import declarative_base
+from sqlalchemy.orm import sessionmaker
+from datetime import datetime
+from config import DATABASE_URL, HISTORY_LIMIT
+
+Base = declarative_base()
+
+
+class Report(Base):
+    __tablename__ = 'reports'
+
+    id = Column(Integer, primary_key=True)
+    report_id = Column(String, unique=True)
+    defect_type = Column(String)
+    description = Column(String)
+    features = Column(JSON)
+    created_at = Column(DateTime, default=datetime.utcnow)
+
+
+engine = create_engine(DATABASE_URL)
+Base.metadata.create_all(engine)
+Session = sessionmaker(bind=engine)
+
+
+def save_report(report):
+    session = Session()
+    new_report = Report(
+        report_id=report['编号'],
+        defect_type=report['缺陷类型'],
+        description=report['缺陷描述'],
+        features=report['特征详情']
+    )
+    session.add(new_report)
+    session.commit()
+    session.close()
+
+
+def get_report_history():
+    session = Session()
+    reports = session.query(Report).order_by(
+        Report.created_at.desc()).limit(HISTORY_LIMIT).all()
+    session.close()
+    return reports

feature_extraction.py

@@ -0,0 +1,66 @@
+import numpy as np
+from scipy import signal
+from config import AMPLITUDE_THRESHOLD, DISTRIBUTION_THRESHOLD, ATTENUATION_THRESHOLD, REFLECTION_THRESHOLD
+
+
+def calculate_amplitude(image):
+    return np.max(image)
+
+
+def classify_amplitude(amplitude):
+    return "强" if amplitude > AMPLITUDE_THRESHOLD else "弱"
+
+
+def calculate_distribution_range(box):
+    width, height = box[2] - box[0], box[3] - box[1]
+    return width * height
+
+
+def classify_distribution_range(area):
+    return "大" if area > DISTRIBUTION_THRESHOLD else "小"
+
+
+def calculate_attenuation_rate(image):
+    gradient = np.gradient(np.array(image).mean(axis=2))
+    return np.mean(np.abs(gradient))
+
+
+def classify_attenuation_rate(rate):
+    return "快" if rate > ATTENUATION_THRESHOLD else "慢"
+
+
+def count_reflections(image, prominence=10):
+    gray = np.mean(np.array(image), axis=2)
+    peaks, _ = signal.find_peaks(np.mean(gray, axis=1), prominence=prominence)
+    return len(peaks)
+
+
+def classify_reflections(count):
+    return "多次反射" if count >= REFLECTION_THRESHOLD else "单次反射"
+
+
+def extract_features(image, detection_result):
+    np_image = np.array(image)
+    amplitude = calculate_amplitude(np_image)
+    amplitude_class = classify_amplitude(amplitude)
+
+    box = detection_result['boxes'][0].tolist()
+    distribution_range = calculate_distribution_range(box)
+    distribution_class = classify_distribution_range(distribution_range)
+
+    attenuation_rate = calculate_attenuation_rate(np_image)
+    attenuation_class = classify_attenuation_rate(attenuation_rate)
+
+    reflection_count = count_reflections(np_image)
+    reflection_class = classify_reflections(reflection_count)
+
+    return {
+        "振幅": amplitude_class,
+        "分布范围": distribution_class,
+        "衰减速度": attenuation_class,
+        "反射次数": reflection_class,
+        "振幅值": amplitude,
+        "分布范围值": distribution_range,
+        "衰减速度值": attenuation_rate,
+        "反射次数值": reflection_count
+    }

model.py

@@ -0,0 +1,22 @@
+from transformers import AutoFeatureExtractor, AutoModelForObjectDetection
+import torch
+from config import MODEL_NAME
+
+
+class RadarDetectionModel:
+    def __init__(self):
+        self.feature_extractor = AutoFeatureExtractor.from_pretrained(
+            MODEL_NAME)
+        self.model = AutoModelForObjectDetection.from_pretrained(MODEL_NAME)
+        self.model.eval()
+
+    @torch.no_grad()
+    def detect(self, image):
+        inputs = self.feature_extractor(images=image, return_tensors="pt")
+        outputs = self.model(**inputs)
+
+        target_sizes = torch.tensor([image.size[::-1]])
+        results = self.feature_extractor.post_process_object_detection(
+            outputs, threshold=0.5, target_sizes=target_sizes)[0]
+
+        return results

report_generation.py

@@ -0,0 +1,31 @@
+from datetime import datetime
+from jinja2 import Environment, FileSystemLoader
+import os
+
+env = Environment(loader=FileSystemLoader('templates'))
+
+
+def generate_report(detection_result, image, features):
+    report_id = f"KD-{datetime.now().strftime('%Y%m%d%H%M%S')}"
+
+    defect_type = "空洞" if features['分布范围'] == "大" or features['反射次数'] == "多次反射" else "裂缝"
+
+    description = (f"{defect_type}，振幅{features['振幅']}，分布范围{features['分布范围']}，"
+                   f"衰减速度{features['衰减速度']}，反射次数{features['反射次数']}")
+
+    report = {
+        "编号": report_id,
+        "缺陷类型": defect_type,
+        "测线位置": "拱顶",  # 假设固定位置
+        "雷达图谱": image,
+        "缺陷描述": description,
+        "验证情况描述": "待验证",
+        "特征详情": features
+    }
+
+    return report
+
+
+def render_report(report):
+    template = env.get_template('report.html')
+    return template.render(report=report)

requirements.txt

@@ -0,0 +1,11 @@
+gradio==3.37.0
+torch==2.0.1
+transformers==4.31.0
+Pillow==9.5.0
+numpy==1.23.5
+matplotlib==3.7.1
+pandas==1.5.3
+sqlalchemy==2.0.19
+plotly==5.15.0
+scikit-learn==1.3.0
+jinja2==3.1.2

static/script.js

@@ -0,0 +1,20 @@
+document.addEventListener('DOMContentLoaded', function () {
+    // 添加交互性功能
+    const tabButtons = document.querySelectorAll('.tab-button')
+    const tabContents = document.querySelectorAll('.tab-content')
+
+    tabButtons.forEach((button) => {
+        button.addEventListener('click', () => {
+            const tabId = button.getAttribute('data-tab')
+
+            tabButtons.forEach((btn) => btn.classList.remove('active'))
+            tabContents.forEach((content) => content.classList.remove('active'))
+
+            button.classList.add('active')
+            document.getElementById(tabId).classList.add('active')
+        })
+    })
+
+    // 添加图表交互性（如果需要）
+    // 这里可以添加与 Plotly 图表交互的代码
+})

static/style.css

@@ -0,0 +1,86 @@
+body {
+    font-family: Arial, sans-serif;
+    line-height: 1.6;
+    color: #333;
+    max-width: 1200px;
+    margin: 0 auto;
+    padding: 20px;
+    background-color: #f4f4f4;
+}
+
+.report-container {
+    background-color: #fff;
+    border-radius: 5px;
+    padding: 20px;
+    margin-bottom: 20px;
+    box-shadow: 0 2px 5px rgba(0,0,0,0.1);
+}
+
+h2 {
+    color: #2c3e50;
+    border-bottom: 2px solid #2c3e50;
+    padding-bottom: 10px;
+}
+
+h3 {
+    color: #34495e;
+}
+
+ul {
+    list-style-type: none;
+    padding-left: 0;
+}
+
+li {
+    margin-bottom: 5px;
+}
+
+.detection-image {
+    max-width: 100%;
+    height: auto;
+    border-radius: 5px;
+    margin-bottom: 20px;
+}
+
+.history-container {
+    background-color: #fff;
+    border-radius: 5px;
+    padding: 20px;
+    margin-top: 20px;
+    box-shadow: 0 2px 5px rgba(0,0,0,0.1);
+}
+
+.history-item {
+    border-bottom: 1px solid #eee;
+    padding: 10px 0;
+}
+
+.history-item:last-child {
+    border-bottom: none;
+}
+
+.feature-chart {
+    width: 100%;
+    height: 400px;
+}
+
+.tabs {
+    margin-bottom: 20px;
+}
+
+.tab-button {
+    background-color: #3498db;
+    color: white;
+    border: none;
+    padding: 10px 20px;
+    cursor: pointer;
+    transition: background-color 0.3s;
+}
+
+.tab-button:hover {
+    background-color: #2980b9;
+}
+
+.tab-button.active {
+    background-color: #2c3e50;
+}

templates/history.html

@@ -0,0 +1,18 @@
+<div class="history-container">
+    <h2>历史记录</h2>
+    {% for report in reports %}
+    <div class="history-item">
+        <p><strong>报告ID:</strong> {{ report.report_id }}</p>
+        <p><strong>缺陷类型:</strong> {{ report.defect_type }}</p>
+        <p><strong>描述:</strong> {{ report.description }}</p>
+        <p><strong>创建时间:</strong> {{ report.created_at }}</p>
+        <h3>特征详情:</h3>
+        <ul>
+            <li>振幅: {{ report.features.振幅 }} ({{ report.features.振幅值|round(2) }})</li>
+            <li>分布范围: {{ report.features.分布范围 }} ({{ report.features.分布范围值|round(2) }})</li>
+            <li>衰减速度: {{ report.features.衰减速度 }} ({{ report.features.衰减速度值|round(2) }})</li>
+            <li>反射次数: {{ report.features.反射次数 }} ({{ report.features.反射次数值 }})</li>
+        </ul>
+    </div>
+    {% endfor %}
+</div>

templates/report.html

@@ -0,0 +1,15 @@
+<div class="report-container">
+    <h2>雷达图谱分析报告</h2>
+    <p><strong>编号:</strong> {{ report.编号 }}</p>
+    <p><strong>缺陷类型:</strong> {{ report.缺陷类型 }}</p>
+    <p><strong>测线位置:</strong> {{ report.测线位置 }}</p>
+    <p><strong>缺陷描述:</strong> {{ report.缺陷描述 }}</p>
+    <p><strong>验证情况描述:</strong> {{ report.验证情况描述 }}</p>
+    <h3>特征详情:</h3>
+    <ul>
+        <li>振幅: {{ report.特征详情.振幅 }} ({{ report.特征详情.振幅值|round(2) }})</li>
+        <li>分布范围: {{ report.特征详情.分布范围 }} ({{ report.特征详情.分布范围值|round(2) }})</li>
+        <li>衰减速度: {{ report.特征详情.衰减速度 }} ({{ report.特征详情.衰减速度值|round(2) }})</li>
+        <li>反射次数: {{ report.特征详情.反射次数 }} ({{ report.特征详情.反射次数值}})</li>
+    </ul>
+</div>

utils.py

@@ -0,0 +1,24 @@
+import matplotlib.pyplot as plt
+import io
+import base64
+
+
+def plot_detection(image, detection_result):
+    plt.figure(figsize=(10, 10))
+    plt.imshow(image)
+    ax = plt.gca()
+
+    for score, label, box in zip(detection_result["scores"], detection_result["labels"], detection_result["boxes"]):
+        x, y, w, h = box
+        rect = plt.Rectangle((x, y), w-x, h-y, fill=False, color='red')
+        ax.add_patch(rect)
+        ax.text(x, y, f'{label}: {score:.2f}',
+                bbox=dict(facecolor='white', alpha=0.8))
+
+    plt.axis('off')
+
+    buf = io.BytesIO()
+    plt.savefig(buf, format='png')
+    buf.seek(0)
+    img_str = base64.b64encode(buf.getvalue()).decode()
+    return f"data:image/png;base64,{img_str}"