https://bugs.kde.org/show_bug.cgi?id=426175
--- Comment #42 from caulier.gil...@gmail.com --- About the Valgrind trace, all unititialized memory entries are like this : ==11721== at 0x56FDAA1: int const& qMin<int>(int const&, int const&) (qglobal.h:594) ==11721== by 0x5EB28CE: Digikam::DNNFaceDetectorBase::correctBbox(cv::Rect_<int>&, cv::Size_<int> const&) const (dnnfacedetectorbase.cpp:127) ==11721== by 0x5EAE704: Digikam::DNNFaceDetectorSSD::postprocess(cv::Mat, cv::Size_<int> const&, std::vector<cv::Rect_<int>, std::allocator<cv::Rect_<int> > >&) const (dnnfacedetectorssd.cpp:144) ==11721== by 0x5EAE18F: Digikam::DNNFaceDetectorSSD::detectFaces(cv::Mat const&, cv::Size_<int> const&, std::vector<cv::Rect_<int>, std::allocator<cv::Rect_<int> > >&) (dnnfacedetectorssd.cpp:78) ==11721== by 0x5EAD201: Digikam::OpenCVDNNFaceDetector::cvDetectFaces(cv::Mat const&, cv::Size_<int> const&) (opencvdnnfacedetector.cpp:274) ==11721== by 0x5EAD09C: Digikam::OpenCVDNNFaceDetector::detectFaces(cv::Mat const&, cv::Size_<int> const&) (opencvdnnfacedetector.cpp:245) ==11721== by 0x5E8C7D5: Digikam::FaceDetector::detectFaces(Digikam::DImg const&, QSize const&) (facedetector.cpp:289) ==11721== by 0x4E2E481: Digikam::DetectionWorker::process(QExplicitlySharedDataPointer<Digikam::FacePipelineExtendedPackage>) (detectionworker.cpp:55) ==11721== by 0x4E0CD4B: Digikam::DetectionWorker::qt_static_metacall(QObject*, QMetaObject::Call, int, void**) (moc_detectionworker.cpp:87) All are relevant of a use of qMax/qMin from Qt in code using OpenCV API. I will use cv::max/cv::min instead, just to be homogeneous, but it's not the source of the real memory leak problem. Gilles -- You are receiving this mail because: You are watching all bug changes.