DeXun/rtsp-video-analysis-system
9
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
f0b4c5a8bfc152ad9657498da06fb4cb0092d00b
rtsp-video-analysis-system/ruoyi-video/src/main/java/com/ruoyi/video/service/VideoAnalysisService.java

457 lines
20 KiB
Java
Raw Normal View History

将检测任务迁移python
2025-09-30 14:23:33 +08:00
package com.ruoyi.video.service;
import com.ruoyi.common.utils.spring.SpringUtils;
import com.ruoyi.framework.service.MinioService;
import com.ruoyi.video.domain.AlarmRecord;
import com.ruoyi.video.domain.Detection;
import com.ruoyi.video.domain.InspectionTask;
import com.ruoyi.video.domain.VMinioObject;
import com.ruoyi.video.mapper.AlarmRecordMapper;
import com.ruoyi.video.mapper.InspectionTaskMapper;
import com.ruoyi.video.thread.detector.HttpYoloDetector;
import com.ruoyi.video.utils.Overlay;
import com.ruoyi.video.utils.CustomMultipartFile;
import lombok.extern.slf4j.Slf4j;
import org.bytedeco.ffmpeg.avcodec.AVPacket;
import org.bytedeco.ffmpeg.global.avcodec;
import org.bytedeco.ffmpeg.global.avutil;
import org.bytedeco.javacv.FFmpegFrameGrabber;
import org.bytedeco.javacv.FFmpegFrameRecorder;
import org.bytedeco.javacv.Frame;
import org.bytedeco.javacv.OpenCVFrameConverter;
import org.bytedeco.opencv.opencv_core.Mat;
import org.bytedeco.opencv.opencv_core.Point;
import org.bytedeco.opencv.opencv_core.Rect;
import org.bytedeco.opencv.opencv_core.Scalar;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.scheduling.annotation.Async;
import org.springframework.stereotype.Service;
import org.springframework.util.CollectionUtils;
import java.io.File;
import java.io.FileOutputStream;
import java.io.InputStream;
import java.nio.IntBuffer;
import java.util.ArrayList;
import java.util.Date;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import java.util.UUID;
import static org.bytedeco.opencv.global.opencv_imgproc.*;
/**
* 视频分析服务 - 离线分析视频并处理结果
*/
@Slf4j
@Service
public class VideoAnalysisService {
@Autowired
private MinioService minioService;
@Autowired
private IVMinioObjectService vMinioObjectService;
@Autowired
private InspectionTaskMapper taskMapper;
@Autowired
private AlarmRecordMapper alarmRecordMapper;
将检测任务迁移python
2025-09-30 17:03:41 +08:00
将检测任务迁移python
2025-09-30 14:23:33 +08:00
@Autowired
private com.ruoyi.video.mapper.InspectionTaskRecordMapper inspectionTaskRecordMapper;
提高视频帧率
2025-10-08 11:51:28 +08:00
// 检测器配置 - 支持环境变量配置
private static final String PYTHON_API_URL = System.getenv().getOrDefault("PYTHON_API_URL", "http://localhost:8000") + "/api/detect/file";
private static final String MODEL_NAME = "smoke"; // 默认使用吸烟检测模型
将检测任务迁移python
2025-09-30 14:23:33 +08:00
/**
* 分析视频并更新记录（同步调用）
* @param task 巡检任务
* @param videoFile 视频文件
*/
public void analyzeVideoWithRecord(InspectionTask task, com.ruoyi.video.domain.InspectionTaskRecord record, File videoFile) {
log.info("开始分析视频并更新记录: 任务ID={}, 记录ID={}", task.getTaskId(), record.getRecordId());
try {
// 创建输出视频文件
File outputVideoFile = File.createTempFile("analysis_output_", ".mp4");
// 创建检测器
feat(video): 添加模型名称字段以支持动态模型选择 - 在 InspectionTask 实体类中新增 modelName 字段及其 getter/setter 方法 - 更新 MyBatis 映射文件，增加对 model_name 字段的映射和支持 - 修改 SQL 查询语句，在查询条件和插入、更新操作中加入 modelName 字段处理 - 调整 VideoAnalysisService 中的模型选择逻辑，优先使用任务配置的模型名称 - 记录日志输出所使用的模型名称及对应的任务ID，便于追踪分析过程
2025-10-07 16:07:23 +08:00
String chosenModel = (task.getModelName() != null && !task.getModelName().trim().isEmpty())
? task.getModelName().trim()
: MODEL_NAME;
log.info("使用模型进行分析: {} (taskId={})", chosenModel, task.getTaskId());
HttpYoloDetector detector = new HttpYoloDetector("yolov8", PYTHON_API_URL, chosenModel, 0x00FF00);
将检测任务迁移python
2025-09-30 14:23:33 +08:00
// 处理视频并记录检测结果
String detectionResult = processVideoWithRecord(videoFile, outputVideoFile, detector, task, record);
// 更新记录的识别结果
record.setResult(detectionResult);
inspectionTaskRecordMapper.updateInspectionTaskRecord(record);
// 清理临时输出文件
if (outputVideoFile.exists()) {
outputVideoFile.delete();
}
log.info("视频分析完成: 任务ID={}, 记录ID={}, 检测结果={}", task.getTaskId(), record.getRecordId(), detectionResult);
} catch (Exception e) {
log.error("视频分析失败: 任务ID={}, 记录ID={}, 错误={}", task.getTaskId(), record.getRecordId(), e.getMessage(), e);
// 更新记录为部分成功
record.setStatus(2);
record.setResult("分析失败: " + e.getMessage());
inspectionTaskRecordMapper.updateInspectionTaskRecord(record);
}
}
/**
* 处理视频并记录检测结果
* @return 检测结果摘要
*/
private String processVideoWithRecord(File inputFile, File outputFile, HttpYoloDetector detector,
InspectionTask task, com.ruoyi.video.domain.InspectionTaskRecord record) throws Exception {
FFmpegFrameGrabber grabber = new FFmpegFrameGrabber(inputFile);
grabber.start();
FFmpegFrameRecorder recorder = new FFmpegFrameRecorder(outputFile,
grabber.getImageWidth(), grabber.getImageHeight(), grabber.getAudioChannels());
recorder.setFormat("mp4");
recorder.setVideoCodec(avcodec.AV_CODEC_ID_H264);
recorder.setFrameRate(grabber.getFrameRate());
recorder.setVideoBitrate(grabber.getVideoBitrate());
if (grabber.getAudioChannels() > 0) {
recorder.setAudioCodec(avcodec.AV_CODEC_ID_AAC);
recorder.setAudioBitrate(grabber.getAudioBitrate());
recorder.setAudioChannels(grabber.getAudioChannels());
recorder.setSampleRate(grabber.getSampleRate());
}
recorder.start();
OpenCVFrameConverter.ToMat converter = new OpenCVFrameConverter.ToMat();
// 用于去重的垃圾检测结果缓存
Map<String, Long> detectedGarbageCache = new HashMap<>();
Map<String, Integer> detectionCounts = new HashMap<>(); // 统计每种类别的数量
// 跟踪检测到的垃圾ID
final Long[] detectionId = {1L};
// 帧计数
long frameCount = 0;
int totalDetections = 0;
Frame frame;
while ((frame = grabber.grab()) != null) {
frameCount++;
if (frame.image != null) {
// 处理视频帧
Mat mat = converter.convert(frame);
if (mat != null && !mat.isNull()) {
// 每10帧执行一次检测，减少API调用频率
if (frameCount % 10 == 0) {
List<Detection> detections = detector.detect(mat);
if (!CollectionUtils.isEmpty(detections)) {
for (Detection detection : detections) {
// 检查是否为新的垃圾检测结果
String detectionKey = generateDetectionKey(detection);
if (!detectedGarbageCache.containsKey(detectionKey)) {
// 这是新检测到的垃圾
detectedGarbageCache.put(detectionKey, detectionId[0]++);
totalDetections++;
// 统计类别数量
String label = detection.getLabel();
detectionCounts.put(label, detectionCounts.getOrDefault(label, 0) + 1);
// 创建告警记录（不重复）
createAlarmRecordForRecord(task, record, detection, mat, frameCount);
} else {
// 更新上次检测时间
detectedGarbageCache.put(detectionKey, detectionId[0]++);
}
}
// 清理超过60秒未检测到的垃圾（假设30fps）
Long currentId = detectionId[0];
detectedGarbageCache.entrySet().removeIf(entry ->
(currentId - entry.getValue()) > grabber.getFrameRate() * 60);
}
}
// 转回Frame并写入录像
Frame processedFrame = converter.convert(mat);
recorder.record(processedFrame);
} else {
// 原样写入
recorder.record(frame);
}
} else if (frame.samples != null) {
// 音频帧原样写入
recorder.record(frame);
}
}
recorder.stop();
recorder.close();
grabber.stop();
grabber.close();
// 上传处理后的视频到MinIO
uploadProcessedVideoForRecord(outputFile, task, record);
// 生成检测结果摘要
StringBuilder resultSummary = new StringBuilder();
resultSummary.append("共检测到 ").append(totalDetections).append(" 个问题");
if (!detectionCounts.isEmpty()) {
resultSummary.append("，详情：");
detectionCounts.forEach((label, count) ->
resultSummary.append(label).append("(").append(count).append(") "));
}
return resultSummary.toString();
}
/**
* 为记录创建告警
*/
private void createAlarmRecordForRecord(InspectionTask task, com.ruoyi.video.domain.InspectionTaskRecord record,
Detection detection, Mat frame, long frameCount) throws Exception {
// 创建告警图像临时文件
File alarmImageFile = File.createTempFile("alarm_", ".jpg");
// 裁剪检测区域，略微扩大区域
Rect rect = detection.getRect();
int x = Math.max(0, rect.x() - 10);
int y = Math.max(0, rect.y() - 10);
int w = Math.min(frame.cols() - x, rect.width() + 20);
int h = Math.min(frame.rows() - y, rect.height() + 20);
// 使用OpenCV保存告警图片
Mat roi = new Mat(frame, new Rect(x, y, w, h));
org.bytedeco.opencv.global.opencv_imgcodecs.imwrite(alarmImageFile.getAbsolutePath(), roi);
// 上传告警图片到MinIO
String fileName = "alarm_" + System.currentTimeMillis() + ".jpg";
String bucketName = "alarm-images";
CustomMultipartFile multipartFile = new CustomMultipartFile(alarmImageFile, fileName, "image/jpeg");
修复工作
2025-10-01 22:19:45 +08:00
String imagePath = minioService.putObject(bucketName, fileName, multipartFile.getInputStream());
将检测任务迁移python
2025-09-30 14:23:33 +08:00
修复工作
2025-10-01 22:19:45 +08:00
// 创建告警记录（严格按照DDL）
将检测任务迁移python
2025-09-30 14:23:33 +08:00
AlarmRecord alarmRecord = new AlarmRecord();
修复工作
2025-10-01 22:19:45 +08:00
alarmRecord.setTaskId(task.getTaskId());
alarmRecord.setTaskName(task.getTaskName());
将检测任务迁移python
2025-09-30 14:23:33 +08:00
alarmRecord.setDeviceId(task.getDeviceId());
修复工作
2025-10-01 22:19:45 +08:00
alarmRecord.setDeviceName(task.getDeviceName());
将检测任务迁移python
2025-09-30 14:23:33 +08:00
alarmRecord.setAlarmType("detection");
修复工作
2025-10-01 22:19:45 +08:00
alarmRecord.setAlarmLevel("2"); // 2=中级
alarmRecord.setAlarmDesc(detection.getLabel() + " - 置信度: " + String.format("%.2f", detection.getConfidence()));
alarmRecord.setConfidence(java.math.BigDecimal.valueOf(detection.getConfidence()));
alarmRecord.setImagePath(imagePath);
alarmRecord.setAlarmTime(new Date());
alarmRecord.setHandleStatus("0"); // 0=未处理
alarmRecord.setCreateBy("system");
将检测任务迁移python
2025-09-30 14:23:33 +08:00
alarmRecord.setCreateTime(new Date());
alarmRecordMapper.insertAlarmRecord(alarmRecord);
log.info("创建告警记录: 类型={}, 任务ID={}, 记录ID={}, 告警ID={}",
detection.getLabel(), task.getTaskId(), record.getRecordId(), alarmRecord.getAlarmId());
// 删除临时文件
if (alarmImageFile.exists()) {
alarmImageFile.delete();
}
}
/**
* 上传处理后的视频（针对记录）
*/
private void uploadProcessedVideoForRecord(File videoFile, InspectionTask task,
com.ruoyi.video.domain.InspectionTaskRecord record) throws Exception {
String fileName = "processed_" + System.currentTimeMillis() + ".mp4";
String bucketName = "inspection-videos";
CustomMultipartFile multipartFile = new CustomMultipartFile(videoFile, fileName, "video/mp4");
String objectUrl = minioService.putObject(bucketName, fileName, multipartFile.getInputStream());
VMinioObject minioObject = new VMinioObject();
minioObject.setBucketName(bucketName);
minioObject.setObjectName(fileName);
修复工作
2025-09-30 18:07:39 +08:00
minioObject.setUrl(objectUrl);
将检测任务迁移python
2025-09-30 14:23:33 +08:00
minioObject.setCreateBy("system");
minioObject.setCreateTime(new Date());
Long objectId = vMinioObjectService.insertVMinioObject(minioObject);
// 更新记录的附件信息，添加处理后的视频URL
String currentAccessory = record.getAccessory();
修复工作
2025-10-01 22:19:45 +08:00
if (currentAccessory == null || currentAccessory.isEmpty()) {
record.setAccessory(objectUrl);
} else {
fix(video): 修改视频附件分隔符 - 将附件URL拼接分隔符从分号改为逗号 - 更新 InspectionTaskRecord 的附件字段更新逻辑- 确保附件列表格式统一，便于后续解析处理
2025-10-07 11:35:53 +08:00
record.setAccessory(currentAccessory + "," + objectUrl);
修复工作
2025-10-01 22:19:45 +08:00
}
将检测任务迁移python
2025-09-30 14:23:33 +08:00
inspectionTaskRecordMapper.updateInspectionTaskRecord(record);
修复工作
2025-10-01 22:19:45 +08:00
log.info("处理后视频已上传: 任务ID={}, 记录ID={}, MinIO对象ID={}, URL={}",
task.getTaskId(), record.getRecordId(), objectId, objectUrl);
将检测任务迁移python
2025-09-30 14:23:33 +08:00
}
/**
修复工作
2025-10-01 22:19:45 +08:00
* 处理视频记录
将检测任务迁移python
2025-09-30 14:23:33 +08:00
* @param inputFile 输入视频文件
* @param outputFile 输出视频文件
* @param detector 检测器
* @param task 巡检任务
修复工作
2025-10-01 22:19:45 +08:00
* @param record 巡检记录
将检测任务迁移python
2025-09-30 14:23:33 +08:00
*/
修复工作
2025-10-01 22:19:45 +08:00
private void processVideoForRecord(File inputFile, File outputFile, HttpYoloDetector detector,
InspectionTask task, com.ruoyi.video.domain.InspectionTaskRecord record) throws Exception {
将检测任务迁移python
2025-09-30 14:23:33 +08:00
FFmpegFrameGrabber grabber = new FFmpegFrameGrabber(inputFile);
grabber.start();
FFmpegFrameRecorder recorder = new FFmpegFrameRecorder(outputFile,
grabber.getImageWidth(), grabber.getImageHeight(), grabber.getAudioChannels());
recorder.setFormat("mp4");
recorder.setVideoCodec(avcodec.AV_CODEC_ID_H264);
recorder.setFrameRate(grabber.getFrameRate());
recorder.setVideoBitrate(grabber.getVideoBitrate());
修复工作
2025-10-01 22:19:45 +08:00
if (grabber.getAudioChannels() > 0) {
recorder.setAudioCodec(avcodec.AV_CODEC_ID_AAC);
recorder.setAudioBitrate(grabber.getAudioBitrate());
recorder.setAudioChannels(grabber.getAudioChannels());
recorder.setSampleRate(grabber.getSampleRate());
}
将检测任务迁移python
2025-09-30 14:23:33 +08:00
recorder.start();
OpenCVFrameConverter.ToMat converter = new OpenCVFrameConverter.ToMat();
// 用于去重的垃圾检测结果缓存
Map<String, Long> detectedGarbageCache = new HashMap<>();
// 跟踪检测到的垃圾ID
修复工作
2025-10-01 22:19:45 +08:00
final Long[] detectionId = {1L};
将检测任务迁移python
2025-09-30 14:23:33 +08:00
// 帧计数
long frameCount = 0;
Frame frame;
while ((frame = grabber.grab()) != null) {
frameCount++;
if (frame.image != null) {
// 处理视频帧
Mat mat = converter.convert(frame);
if (mat != null && !mat.isNull()) {
// 每10帧执行一次检测，减少API调用频率
if (frameCount % 10 == 0) {
List<Detection> detections = detector.detect(mat);
if (!CollectionUtils.isEmpty(detections)) {
// 绘制检测框
for (Detection detection : detections) {
drawDetection(mat, detection);
// 检查是否为新的垃圾检测结果
String detectionKey = generateDetectionKey(detection);
if (!detectedGarbageCache.containsKey(detectionKey)) {
// 这是新检测到的垃圾
detectedGarbageCache.put(detectionKey, detectionId[0]++);
修复工作
2025-10-01 22:19:45 +08:00
// 创建告警记录（关联到记录）
createAlarmRecordForRecord(task, record, detection, mat, frameCount);
将检测任务迁移python
2025-09-30 14:23:33 +08:00
} else {
// 更新上次检测时间
detectedGarbageCache.put(detectionKey, detectionId[0]++);
}
}
// 清理超过60秒未检测到的垃圾（假设30fps）
Long currentId = detectionId[0];
detectedGarbageCache.entrySet().removeIf(entry ->
(currentId - entry.getValue()) > grabber.getFrameRate() * 60);
}
}
// 转回Frame并写入录像
Frame processedFrame = converter.convert(mat);
recorder.record(processedFrame);
} else {
// 原样写入
recorder.record(frame);
}
} else if (frame.samples != null) {
// 音频帧原样写入
recorder.record(frame);
}
}
recorder.stop();
recorder.close();
grabber.stop();
grabber.close();
修复工作
2025-10-01 22:19:45 +08:00
// 上传处理后的视频（更新记录）
uploadProcessedVideoForRecord(outputFile, task, record);
将检测任务迁移python
2025-09-30 14:23:33 +08:00
}
修复工作
2025-10-01 22:19:45 +08:00
将检测任务迁移python
2025-09-30 14:23:33 +08:00
/**
* 生成检测结果的唯一键，用于检测结果去重
*/
private String generateDetectionKey(Detection detection) {
// 使用检测框的位置和大小来生成键
// 允许小范围波动，认为是同一个物体
Rect rect = detection.getRect();
int x = rect.x() / 10 * 10; // 取10的倍数，允许小波动
int y = rect.y() / 10 * 10;
int w = rect.width() / 10 * 10;
int h = rect.height() / 10 * 10;
return String.format("%s_%d_%d_%d_%d", detection.getLabel(), x, y, w, h);
}
修复工作
2025-10-01 22:19:45 +08:00
将检测任务迁移python
2025-09-30 14:23:33 +08:00
/**
* 在图像上绘制检测框
*/
private void drawDetection(Mat frame, Detection detection) {
// 绘制边界框
Rect rect = detection.getRect();
Scalar color = new Scalar(detection.getColorBGR());
int thickness = 2;
rectangle(frame, rect, color, thickness, LINE_8, 0);
// 绘制标签背景
String label = String.format("%s: %.2f", detection.getLabel(), detection.getConfidence());
Point textPosition = new Point(rect.x(), rect.y() - 10);
// 获取文本大小 - 添加缺失的IntBuffer参数
IntBuffer baseline = IntBuffer.allocate(1);
org.bytedeco.opencv.opencv_core.Size textSize = getTextSize(
label, FONT_HERSHEY_SIMPLEX, 0.5, thickness, baseline);
// 绘制标签背景矩形
rectangle(frame,
new Rect(rect.x(), rect.y() - textSize.height() - 10,
textSize.width(), textSize.height() + 10),
color, FILLED, LINE_8, 0);
// 绘制文本
putText(frame, label, textPosition,
FONT_HERSHEY_SIMPLEX, 0.5, new Scalar(255, 255, 255, 0),
1, LINE_AA, false);
}
}
Reference in New Issue Copy Permalink