Weekly appimage segfaulted while upgrading faces database, segfaults every time

[Michael Miller]

Hi Michael,
If digiKam segfaulted while retraining, then the retraining isn’t complete.  I’ll change that so it doesn’t mark it as complete until it finishes.

To redo the training, go to Tools->Maintenance->Detect and recognize faces, and select “Rebuild all training data”.  Let that complete before you try any other face options.

Do you have “Use full CPU” turned on in the People sidebar and in the Maintenance view?

Cheers,
Mike

> On Feb 5, 2025, at 10:35 AM, Michael Moore <stuporglue at gmail.com> wrote:
> 
> Hello,
> 
> I have been using 8.5 appimage and decided to try the weekly appimage. I'm using the debug image (digiKam-8.6.0-20250205T150105-Qt6-x86-64-debug.appimage)
> 
> It showed a dialog saying that the Face Recognition database was out of date and results wouldn't be accurate until it was upgraded. I said yes to the upgrade and then allowed the next dialog to download the 4 new image recognition libraries.
> 
> A few moments later Digikam segfaulted.
> 
> When I start it again the database-upgrade notice doesn't return, but I don't know that it completed successfully.
> 
> Q1 - Do I need to trigger something so that the face database upgrade is complete?
> 
> After the segfault Digikam prompted to download the libraries again, but this time those downloaded successfully.
> 
> Now when I try to run facial recognition I get a segfault every time. I set the dialog to "Recognize faces only" and then click "Scan collection for faces". 
> 
> Q2 - Is this a known / expected segfault, or should I collect some more info?
> 
> The last lines of output in the terminal look like this.
> 
> Digikam::DImg::load: "/shared/big/no_backup/ro_photos/photos/2022/02/18/YFEZ0700.HEIC" : "HEIF" file identified
> Digikam::DMetadata::load: Loading metadata with "Exiv2" backend from "/shared/big/no_backup/ro_photos/photos/2022/02/18/YFEZ0700.HEIC"
> unknown: HEIF color profile found with size: 548
> unknown: HEIF image size: ( 3024 x 4032 )
> unknown: Decoded HEIF image properties: size( 4032 x 3024 ), Alpha: false , Color depth : 8
> unknown: HEIF data container: 0x7f75e5d1d010
> unknown: HEIC bytes per line: 12096
> unknown: Color bytes depth: 8
> unknown: Color multiplier: 1
> Digikam::StackedView::setViewMode: Stacked View Mode :  1
> Digikam::FocusPointGroup::load: FocusPointsGroup: no Point to load
> Digikam::PreviewLoadingTask::execute: Try to get preview from "/shared/big/no_backup/ro_photos/photos/2008/09/04/yellowstone blog - 07.jpg"
> Digikam::PreviewLoadingTask::execute: Preview quality:  0
> Digikam::DMetadata::load: Loading metadata with "Exiv2" backend from "/shared/big/no_backup/ro_photos/photos/2008/09/04/yellowstone blog - 07.jpg"
> Digikam::PreviewLoadingTask::loadImagePreview: Try to load DImg preview from: "/shared/big/no_backup/ro_photos/photos/2008/09/04/yellowstone blog - 07.jpg"
> Digikam::DImg::load: "/shared/big/no_backup/ro_photos/photos/2008/09/04/yellowstone blog - 07.jpg" : "JPEG" file identified
> Digikam::DMetadata::load: Loading metadata with "Exiv2" backend from "/shared/big/no_backup/ro_photos/photos/2008/09/04/yellowstone blog - 07.jpg"
> Digikam::FaceDb::setting: FaceDB SelectFaceSetting val ret = 0
> Digikam::FaceDb::setting: FaceDB SelectFaceSetting val ret = 0
> Digikam::FacesEngine::slotStart: Total is 28029
> Digikam::DNNFaceDetectorYuNet::DNNFaceDetectorYuNet: Creating new instance of DNNFaceDetectorYuNet
> Digikam::DNNFaceDetectorYuNet::loadModels: Face detection model: "YuNet" ready
> Digikam::MLPipelineFoundation::checkMoreWorkers: MLPipelineFoundation::checkMoreWorkers: Sending signal to more workers
> Digikam::MLPipelineFoundation::checkMoreWorkers: MLPipelineFoundation::checkMoreWorkers: Sending signal to more workers
> Digikam::MLPipelineFoundation::checkMoreWorkers: MLPipelineFoundation::checkMoreWorkers: Sending signal to more workers
> [ WARN:0 at 80.706] global ocl4dnn_conv_spatial.cpp:1931 loadTunedConfig OpenCV(ocl4dnn): consider to specify kernel configuration cache directory through OPENCV_OCL4DNN_CONFIG_PATH p
> arameter.
> OpenCL program build log: dnn/dummy
> Status -11: CL_BUILD_PROGRAM_FAILURE
> -cl-no-subgroup-ifp
> Error in processing command line: Don't understand command line argument "-cl-no-subgroup-ifp"!
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Finish aligning face in  98  ms
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Start neural network
> Digikam::DNNSFaceExtractor::getFaceEmbedding: No face landmarks found
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Finish computing face embedding in  106  ms
> Digikam::DNNSFaceExtractor::getFaceEmbedding: No face landmarks found
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Finish computing face embedding in  122  ms
> Digikam::DNNSFaceExtractor::getFaceEmbedding: No face landmarks found
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Finish computing face embedding in  142  ms
> Digikam::DNNSFaceExtractor::getFaceEmbedding: No face landmarks found
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Finish computing face embedding in  161  ms
> Digikam::DNNSFaceExtractor::getFaceEmbedding: No face landmarks found
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Finish computing face embedding in  104  ms
> Digikam::DNNSFaceExtractor::getFaceEmbedding: No face landmarks found
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Finish computing face embedding in  115  ms
> Digikam::DNNSFaceExtractor::getFaceEmbedding: No face landmarks found
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Finish computing face embedding in  72  ms
> Digikam::DNNSFaceExtractor::getFaceEmbedding: No face landmarks found
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Finish computing face embedding in  49  ms
> Digikam::DNNSFaceExtractor::getFaceEmbedding: No face landmarks found
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Finish computing face embedding in  56  ms
> Digikam::DNNSFaceExtractor::getFaceEmbedding: No face landmarks found
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Finish computing face embedding in  13  ms
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Finish computing face embedding in  325  ms
> Digikam::FaceClassifier::predictFullSearch: FaceClassifier::predictFullSearch: classifier prediction is:  159  completed in  0
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Finish aligning face in  426  ms
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Start neural network
> Digikam::DNNSFaceExtractor::getFaceEmbedding: No face landmarks found
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Finish computing face embedding in  68  ms
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Finish computing face embedding in  105  ms
> Digikam::FaceClassifier::predictFullSearch: FaceClassifier::predictFullSearch: classifier prediction is:  -1  completed in  0
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Finish aligning face in  532  ms
> Digikam::DNNSFaceExtractor::getFaceEmbedding: Start neural network
> /tmp/user/1001/.mount_digiKaO6gtZg/AppRun: line 261: 149464 Segmentation fault      (core dumped) digikam $@
> 
> It says core dumped, but I don't see the dump file.
> 
> Thanks,
> Michael