Skip to content

Commit

Permalink
Merge branch 'master' of https://github.com/Nuzhny007/Multitarget-tra…
Browse files Browse the repository at this point in the history 
…cker
  • Loading branch information
@Nuzhny007
Nuzhny007 committed Apr 28, 2023
2 parents 5ee4fcb + 581ae98 commit 5d675b9
Show file tree
Hide file tree
Showing 25 changed files with 2,481 additions and 212 deletions.
1 change: 0 additions & 1 deletion async_detector/AsyncDetector.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -392,7 +392,6 @@ void AsyncDetector::DetectThread(const config_t& config, cv::Mat firstFrame, Fra
{
cv::UMat ufirst = firstFrame.getUMat(cv::ACCESS_READ);
std::unique_ptr<BaseDetector> detector = BaseDetector::CreateDetector(tracking::Detectors::Yolo_Darknet, config, ufirst);
detector->SetMinObjectSize(cv::Size(firstFrame.cols / 50, firstFrame.cols / 50));

for (; !(*stopFlag);)
{
Expand Down
127 changes: 111 additions & 16 deletions combined/combined.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -190,7 +190,7 @@ void CombinedDetector::DetectAndTrack(cv::Mat frame)
{
if (track.m_isStatic)
{
//m_detectorBGFG->UpdateIgnoreMask(uGray, track.m_rrect.boundingRect());
m_detectorBGFG->UpdateIgnoreMask(uGray, track.m_rrect.boundingRect());
}
}

Expand All @@ -199,6 +199,7 @@ void CombinedDetector::DetectAndTrack(cv::Mat frame)
const regions_t& regsBGFG = m_detectorBGFG->GetDetects();

m_trackerBGFG->Update(regsBGFG, uGray, m_fps);
m_trackerBGFG->Update(regions_t(), uGray, m_fps);

m_trackerBGFG->GetTracks(m_tracksBGFG);
for (const auto& bbox : m_oldBoxes)
Expand Down Expand Up @@ -314,45 +315,122 @@ bool CombinedDetector::InitDetector(cv::UMat frame)
#else
std::string pathToModel = "../data/";
#endif
size_t maxBatch = 1;
enum class YOLOModels
{
TinyYOLOv3 = 0,
YOLOv3,
YOLOv4
YOLOv4,
TinyYOLOv4,
YOLOv5,
YOLOv6,
YOLOv7,
YOLOv7Mask,
YOLOv8
};
YOLOModels usedModel = YOLOModels::YOLOv4;
YOLOModels usedModel = YOLOModels::YOLOv7Mask;
switch (usedModel)
{
case YOLOModels::TinyYOLOv3:
configDNN.emplace("modelConfiguration", pathToModel + "yolov3-tiny.cfg");
configDNN.emplace("modelBinary", pathToModel + "yolov3-tiny.weights");
configDNN.emplace("confidenceThreshold", "0.5");
configDNN.emplace("inference_precision", "FP32");
configDNN.emplace("net_type", "YOLOV3");
maxBatch = 4;
configDNN.emplace("maxCropRatio", "2");
break;

case YOLOModels::YOLOv3:
configDNN.emplace("modelConfiguration", pathToModel + "yolov3.cfg");
configDNN.emplace("modelBinary", pathToModel + "yolov3.weights");
configDNN.emplace("confidenceThreshold", "0.7");
configDNN.emplace("inference_precision", "FP32");
configDNN.emplace("net_type", "YOLOV3");
maxBatch = 2;
configDNN.emplace("maxCropRatio", "-1");
break;

case YOLOModels::YOLOv4:
configDNN.emplace("modelConfiguration", pathToModel + "yolov4.cfg");
configDNN.emplace("modelBinary", pathToModel + "yolov4.weights");
configDNN.emplace("confidenceThreshold", "0.4");
configDNN.emplace("inference_precision", "FP32");
configDNN.emplace("net_type", "YOLOV4");
maxBatch = 1;
configDNN.emplace("maxCropRatio", "-1");
break;

case YOLOModels::TinyYOLOv4:
configDNN.emplace("modelConfiguration", pathToModel + "yolov4-tiny.cfg");
configDNN.emplace("modelBinary", pathToModel + "yolov4-tiny.weights");
configDNN.emplace("confidenceThreshold", "0.5");
configDNN.emplace("inference_precision", "FP32");
configDNN.emplace("net_type", "YOLOV4_TINY");
maxBatch = 1;
configDNN.emplace("maxCropRatio", "-1");
break;

case YOLOModels::YOLOv5:
configDNN.emplace("modelConfiguration", pathToModel + "yolov5s.cfg");
configDNN.emplace("modelBinary", pathToModel + "yolov5s.weights");
configDNN.emplace("confidenceThreshold", "0.5");
configDNN.emplace("inference_precision", "FP32");
configDNN.emplace("net_type", "YOLOV5");
maxBatch = 1;
configDNN.emplace("maxCropRatio", "-1");
break;

case YOLOModels::YOLOv6:
configDNN.emplace("modelConfiguration", pathToModel + "yolov6s.onnx");
configDNN.emplace("modelBinary", pathToModel + "yolov6s.onnx");
configDNN.emplace("confidenceThreshold", "0.5");
configDNN.emplace("inference_precision", "FP32");
configDNN.emplace("net_type", "YOLOV6");
maxBatch = 1;
configDNN.emplace("maxCropRatio", "-1");
break;

case YOLOModels::YOLOv7:
configDNN.emplace("modelConfiguration", pathToModel + "yolov7.onnx");
configDNN.emplace("modelBinary", pathToModel + "yolov7.onnx");
configDNN.emplace("confidenceThreshold", "0.2");
configDNN.emplace("inference_precision", "FP32");
configDNN.emplace("net_type", "YOLOV7");
maxBatch = 1;
configDNN.emplace("maxCropRatio", "-1");
break;

case YOLOModels::YOLOv7Mask:
configDNN.emplace("modelConfiguration", pathToModel + "yolov7-seg_orig.onnx");
configDNN.emplace("modelBinary", pathToModel + "yolov7-seg_orig.onnx");
configDNN.emplace("confidenceThreshold", "0.2");
configDNN.emplace("inference_precision", "FP16");
configDNN.emplace("net_type", "YOLOV7Mask");
maxBatch = 1;
configDNN.emplace("maxCropRatio", "-1");
break;

case YOLOModels::YOLOv8:
configDNN.emplace("modelConfiguration", pathToModel + "yolov8s.onnx");
configDNN.emplace("modelBinary", pathToModel + "yolov8s.onnx");
configDNN.emplace("confidenceThreshold", "0.2");
configDNN.emplace("inference_precision", "FP32");
configDNN.emplace("net_type", "YOLOV8");
maxBatch = 1;
configDNN.emplace("maxCropRatio", "-1");
break;
}
configDNN.emplace("maxBatch", std::to_string(maxBatch));
configDNN.emplace("classNames", pathToModel + "coco.names");
configDNN.emplace("maxCropRatio", "-1");

configDNN.emplace("white_list", "person");
configDNN.emplace("white_list", "car");
configDNN.emplace("white_list", "bicycle");
configDNN.emplace("white_list", "motorbike");
configDNN.emplace("white_list", "bus");
configDNN.emplace("white_list", "truck");
configDNN.emplace("white_list", "vehicle");

m_detectorDNN = BaseDetector::CreateDetector(tracking::Detectors::Yolo_Darknet, configDNN, frame);
configDNN.emplace("white_list", "backpack");
configDNN.emplace("white_list", "handbag");
configDNN.emplace("white_list", "suitcase");

m_detectorDNN = BaseDetector::CreateDetector(tracking::Detectors::Yolo_TensorRT, configDNN, frame);
if (m_detectorDNN.get())
m_detectorDNN->SetMinObjectSize(cv::Size(frame.cols / 40, frame.rows / 40));

Expand Down Expand Up @@ -416,10 +494,24 @@ bool CombinedDetector::InitTracker(cv::UMat frame)
settingsDNN.m_minAreaRadiusPix = -1.f;
#endif
settingsDNN.m_minAreaRadiusK = 0.8f;
settingsDNN.m_maximumAllowedSkippedFrames = cvRound(2 * m_fps); // Maximum allowed skipped frames
settingsDNN.m_maxTraceLength = cvRound(5 * m_fps); // Maximum trace length

//settingsDNN.AddNearTypes("person", "motorbike", true);
settingsDNN.m_useAbandonedDetection = true;
if (settingsDNN.m_useAbandonedDetection)
{
settingsDNN.m_minStaticTime = m_minStaticTime;
settingsDNN.m_maxStaticTime = 30;
settingsDNN.m_maximumAllowedSkippedFrames = cvRound(settingsDNN.m_minStaticTime * m_fps); // Maximum allowed skipped frames
settingsDNN.m_maxTraceLength = 2 * settingsDNN.m_maximumAllowedSkippedFrames; // Maximum trace length
}
else
{
settingsDNN.m_maximumAllowedSkippedFrames = cvRound(2 * m_fps); // Maximum allowed skipped frames
settingsDNN.m_maxTraceLength = cvRound(4 * m_fps); // Maximum trace length
}

settingsDNN.AddNearTypes(TypeConverter::Str2Type("backpack"), TypeConverter::Str2Type("handbag"), true);
settingsDNN.AddNearTypes(TypeConverter::Str2Type("backpack"), TypeConverter::Str2Type("suitcase"), true);
settingsDNN.AddNearTypes(TypeConverter::Str2Type("suitcase"), TypeConverter::Str2Type("handbag"), true);

m_trackerDNN = BaseTracker::CreateTracker(settingsDNN);

Expand Down Expand Up @@ -484,7 +576,8 @@ void CombinedDetector::DrawData(cv::Mat frame, int framesCounter, int currTime)
}
}

m_detectorBGFG->CalcMotionMap(frame);
//m_detectorBGFG->CalcMotionMap(frame);
m_detectorDNN->CalcMotionMap(frame);

for (const auto& track : m_tracksDNN)
{
Expand All @@ -495,6 +588,8 @@ void CombinedDetector::DrawData(cv::Mat frame, int framesCounter, int currTime)
DrawTrack(frame, 1, track);

std::stringstream label;
if (track.m_isStatic)
label << "abandoned ";
label << TypeConverter::Type2Str(track.m_type) << " " << track.m_ID.ID2Str();// << "): " << std::fixed << std::setw(2) << std::setprecision(2) << track.m_confidence;
int baseLine = 0;
cv::Size labelSize = cv::getTextSize(label.str(), cv::FONT_HERSHEY_SIMPLEX, 0.5, 1, &baseLine);
Expand All @@ -520,7 +615,7 @@ void CombinedDetector::DrawData(cv::Mat frame, int framesCounter, int currTime)
brect.y = std::max(0, frame.rows - brect.height - 1);
brect.height = std::min(brect.height, frame.rows - 1);
}
DrawFilledRect(frame, cv::Rect(cv::Point(brect.x, brect.y - labelSize.height), cv::Size(labelSize.width, labelSize.height + baseLine)), cv::Scalar(200, 200, 200), 150);
DrawFilledRect(frame, cv::Rect(cv::Point(brect.x, brect.y - labelSize.height), cv::Size(labelSize.width, labelSize.height + baseLine)), track.m_isStatic ? cv::Scalar(200, 0, 200) : cv::Scalar(200, 200, 200), 150);
cv::putText(frame, label.str(), brect.tl(), cv::FONT_HERSHEY_SIMPLEX, 0.5, cv::Scalar(0, 0, 0));
}
}
Expand Down
4 changes: 3 additions & 1 deletion example/VideoExample.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -586,7 +586,9 @@ void VideoExample::DrawTrack(cv::Mat frame,
cv::Rect brect = track.m_rrect.boundingRect();
std::string label = track.m_ID.ID2Str();
if (track.m_type != bad_type)
label += " (" + TypeConverter::Type2Str(track.m_type) + ")";
label += ": " + TypeConverter::Type2Str(track.m_type);
if (track.m_confidence > 0)
label += ", " + std::to_string(track.m_confidence);
#if 0
track_t mean = 0;
track_t stddev = 0;
Expand Down
Loading

0 comments on commit 5d675b9

Please sign in to comment.