preparing 6.0.0 code...
Locked
 
|
Hi all,
In first, i hope an happy new year to all. 1/ I currently revival the face management git branch written while summer 2017 by the student working on AI engine to recognize faces. The code is sync with master, but do not compile yet. I will fix it. 2/ Important point : this code do not compile with OpenCV 2.x. It require OpenCV 3.1. This want mean that digiKam 6.0.0 will drop OpenCV 2.x support definitively. 3/ when faces management branch will be fixed, I would to create a new one dedicated to 6.0.0 release and let's master for the 5.x series, as certainly a new 5.9.0 will be published few weeks ago 5.8.0, as maintenance, until 6.0.0 code will be stabilized. 4/ The clone tool : this one still not very unusable for 5.8.0. I recommend to disable this option in editor for the moment, until zoom support and brushes size preview over canvas is not implemented. 5/ I do know yet when 5.8.0 will be released, as i don't receive a new response from Nicolas to create the whole tarball. Also Maik, want certainly receive more feedback about Mysql support as i can see recently in bugzilla. Best Gilles Caulier |
|
|
Gilles, and everyone else a Happy New Year. Yes, I'm looking forward to
feedback on MySQL - but only positive. ((:-)) No, seriously, I know that there will be problems. Currently I only use MySQL for testing and it is running really well here. It is important that we write in the release note that the core DB with the migration tool should be copied into a new DB and should be started with a new created thumbnail and face DB. Maik Am Mittwoch, 3. Januar 2018, 18:10:50 CET schrieb Gilles Caulier: > Hi all, > > In first, i hope an happy new year to all. > > 1/ I currently revival the face management git branch written while summer > 2017 by the student working on AI engine to recognize faces. The code is > sync with master, but do not compile yet. I will fix it. > > 2/ Important point : this code do not compile with OpenCV 2.x. It require > OpenCV 3.1. This want mean that digiKam 6.0.0 will drop OpenCV 2.x support > definitively. > > 3/ when faces management branch will be fixed, I would to create a new one > dedicated to 6.0.0 release and let's master for the 5.x series, as > certainly a new 5.9.0 will be published few weeks ago 5.8.0, as > maintenance, until 6.0.0 code will be stabilized. > > 4/ The clone tool : this one still not very unusable for 5.8.0. I recommend > to disable this option in editor for the moment, until zoom support and > brushes size preview over canvas is not implemented. > > 5/ I do know yet when 5.8.0 will be released, as i don't receive a new > response from Nicolas to create the whole tarball. Also Maik, want > certainly receive more feedback about Mysql support as i can see recently > in bugzilla. > > Best > > Gilles Caulier |
|
|
The branch gsoc17-face-mngmnt is now sync with master and compile and link properly. I will create a new 6.0.0 branch based on the gsoc17-face-mngmnt branch for future main development. Gilles Caulier 2018-01-03 19:57 GMT+01:00 Maik Qualmann <[hidden email]>: Gilles, and everyone else a Happy New Year. Yes, I'm looking forward to |
|
|
The 6.0.0 git branch is created in remote : [gilles@localhost digikam.master]$ git branch -a * master remotes/origin/HEAD -> origin/master remotes/origin/development/0.10 remotes/origin/development/0.8 remotes/origin/development/0.8.x remotes/origin/development/0.9.x remotes/origin/development/1.2 remotes/origin/development/1.3 remotes/origin/development/3.0.0 remotes/origin/development/6.0.0 < ============================= remotes/origin/development/balooport remotes/origin/development/bug309058 remotes/origin/development/cmakecleanup remotes/origin/development/externaltools remotes/origin/development/frameworks remotes/origin/development/garbagecollection remotes/origin/development/hupnp remotes/origin/development/imagemetadatacache remotes/origin/development/kde3 remotes/origin/development/photolayoutseditor remotes/origin/development/sql2.0 remotes/origin/development/usbimport remotes/origin/gsoc11-clone remotes/origin/gsoc11-presentation remotes/origin/gsoc13-editorcanvas remotes/origin/gsoc13-imgqsort remotes/origin/gsoc13-localadj remotes/origin/gsoc13-tagmngr remotes/origin/gsoc13-thumbbar-mv remotes/origin/gsoc13-videow remotes/origin/gsoc14-greyc remotes/origin/gsoc14-kipibqm remotes/origin/gsoc14-qaccess remotes/origin/gsoc15-advmetadata remotes/origin/gsoc15-kio-drop remotes/origin/gsoc16-mysql remotes/origin/gsoc16-red-eye remotes/origin/gsoc17-db-sep remotes/origin/gsoc17-dlna remotes/origin/gsoc17-face-mngmnt remotes/origin/gsoc17-healing remotes/origin/master This one is based on gsoc17-face-mngmnt sync with master today morning. The remotes/origin/development/6.0.0 must be used to put next 6.0.0 new code and must be sync with master until 5.x serie will be completed. Best Gilles Caulier 2018-01-04 12:24 GMT+01:00 Gilles Caulier <[hidden email]>:
|
|
|
The main integration and improvements of new face recognition algorithms in 6.0.0 branch is now complete and ready to use. The default algorithm is DNN (deep neural network => deep learning => IA). In face scan dialog, in advanced tab, we have a new option where recognition algorithm can be selected. - lpbh : older one present on 5.x based on OpenCV (https://docs.opencv.org/2.4/modules/contrib/doc/facerec/facerec_tutorial.html#local-binary-patterns-histograms) - fisherfaces : OpenCV based (https://docs.opencv.org/2.4/modules/contrib/doc/facerec/facerec_tutorial.html#fisherfaces) - eigenfaces : OpenCV based (https://docs.opencv.org/2.4/modules/contrib/doc/facerec/facerec_tutorial.html#eigenfaces) - dnn : dlib based (http://blog.dlib.net/2017/02/high-quality-face-recognition-with-deep.html) I'm not yet satisfied by DNN. I'm not started to compare the algorithm with a clean database. In all cases, we must detect all faces, not tagged. After that, we need to train the faces with names, else the recognition will not work. As i know, at least with lpbh algorithm, we need at least 6 images tagged with same name to see a recognition result while scanning. To have a clean compare result, for each algorithm to use, we need a clear database. The student working on this project while summer 2017 report to me that in all cases the best algorithms are : 1 DNN. 2 eigenfaces. 3 fisherfaces. 4 lpbh. The goal is to use the best one at end. Gilles Caulier 2018-01-04 13:18 GMT+01:00 Gilles Caulier <[hidden email]>:
|
|
|
Compile error:
/home/maik/Devel/digikam-software-compilation/core/libs/facesengine/recognition-opencv-eigenfaces/facerec_eigenborrowed.cpp: In static member function ‘static cv::P
Am Samstag, 6. Januar 2018, 17:53:30 CET schrieb Gilles Caulier: > The main integration and improvements of new face recognition algorithms in > 6.0.0 branch is now complete and ready to use. > > The default algorithm is DNN (deep neural network => deep learning => IA). > > In face scan dialog, in advanced tab, we have a new option where > recognition algorithm can be selected. > > - lpbh : older one present on 5.x based on OpenCV ( > https://docs.opencv.org/2.4/modules/contrib/doc/facerec/facerec_tutorial.htm > l#local-binary-patterns-histograms ) > - fisherfaces : OpenCV based ( > https://docs.opencv.org/2.4/modules/contrib/doc/facerec/facerec_tutorial.htm > l#fisherfaces ) > - eigenfaces : OpenCV based ( > https://docs.opencv.org/2.4/modules/contrib/doc/facerec/facerec_tutorial.htm > l#eigenfaces ) > - dnn : dlib based ( > http://blog.dlib.net/2017/02/high-quality-face-recognition-with-deep.html) > > I'm not yet satisfied by DNN. I'm not started to compare the algorithm with > a clean database. In all cases, we must detect all faces, not tagged. After > that, we need to train the faces with names, else the recognition will not > work. As i know, at least with lpbh algorithm, we need at least 6 images > tagged with same name to see a recognition result while scanning. To have a > clean compare result, for each algorithm to use, we need a clear database. > > The student working on this project while summer 2017 report to me that in > all cases the best algorithms are : > > 1 DNN. > 2 eigenfaces. > 3 fisherfaces. > 4 lpbh. > > The goal is to use the best one at end. > > Gilles Caulier > > 2018-01-04 13:18 GMT+01:00 Gilles Caulier <[hidden email]>: > > The 6.0.0 git branch is created in remote : > > > > [gilles@localhost digikam.master]$ git branch -a > > * master > > > > remotes/origin/HEAD -> origin/master > > remotes/origin/development/0.10 > > remotes/origin/development/0.8 > > remotes/origin/development/0.8.x > > remotes/origin/development/0.9.x > > remotes/origin/development/1.2 > > remotes/origin/development/1.3 > > remotes/origin/development/3.0.0 > > remotes/origin/development/6.0.0 < > > > > ============================= > > > > remotes/origin/development/balooport > > remotes/origin/development/bug309058 > > remotes/origin/development/cmakecleanup > > remotes/origin/development/externaltools > > remotes/origin/development/frameworks > > remotes/origin/development/garbagecollection > > remotes/origin/development/hupnp > > remotes/origin/development/imagemetadatacache > > remotes/origin/development/kde3 > > remotes/origin/development/photolayoutseditor > > remotes/origin/development/sql2.0 > > remotes/origin/development/usbimport > > remotes/origin/gsoc11-clone > > remotes/origin/gsoc11-presentation > > remotes/origin/gsoc13-editorcanvas > > remotes/origin/gsoc13-imgqsort > > remotes/origin/gsoc13-localadj > > remotes/origin/gsoc13-tagmngr > > remotes/origin/gsoc13-thumbbar-mv > > remotes/origin/gsoc13-videow > > remotes/origin/gsoc14-greyc > > remotes/origin/gsoc14-kipibqm > > remotes/origin/gsoc14-qaccess > > remotes/origin/gsoc15-advmetadata > > remotes/origin/gsoc15-kio-drop > > remotes/origin/gsoc16-mysql > > remotes/origin/gsoc16-red-eye > > remotes/origin/gsoc17-db-sep > > remotes/origin/gsoc17-dlna > > remotes/origin/gsoc17-face-mngmnt > > remotes/origin/gsoc17-healing > > remotes/origin/master > > > > This one is based on gsoc17-face-mngmnt sync with master today morning. > > > > The remotes/origin/development/6.0.0 must be used to put next 6.0.0 new > > code and must be sync with master until 5.x serie will be completed. > > > > Best > > > > Gilles Caulier > > > > 2018-01-04 12:24 GMT+01:00 Gilles Caulier <[hidden email]>: > >> The branch gsoc17-face-mngmnt is now sync with master and compile and > >> link properly. > >> > >> I will create a new 6.0.0 branch based on the gsoc17-face-mngmnt branch > >> for future main development. > >> > >> Gilles Caulier > >> > >> 2018-01-03 19:57 GMT+01:00 Maik Qualmann <[hidden email]>: > >>> Gilles, and everyone else a Happy New Year. Yes, I'm looking forward to > >>> feedback on MySQL - but only positive. ((:-)) No, seriously, I know that > >>> there > >>> will be problems. Currently I only use MySQL for testing and it is > >>> running > >>> really well here. It is important that we write in the release note that > >>> the > >>> core DB with the migration tool should be copied into a new DB and > >>> should be > >>> started with a new created thumbnail and face DB. > >>> > >>> Maik > >>> > >>> Am Mittwoch, 3. Januar 2018, 18:10:50 CET schrieb Gilles Caulier: > >>> > Hi all, > >>> > > >>> > In first, i hope an happy new year to all. > >>> > > >>> > 1/ I currently revival the face management git branch written while > >>> > >>> summer > >>> > >>> > 2017 by the student working on AI engine to recognize faces. The code > >>> > >>> is > >>> > >>> > sync with master, but do not compile yet. I will fix it. > >>> > > >>> > 2/ Important point : this code do not compile with OpenCV 2.x. It > >>> > >>> require > >>> > >>> > OpenCV 3.1. This want mean that digiKam 6.0.0 will drop OpenCV 2.x > >>> > >>> support > >>> > >>> > definitively. > >>> > > >>> > 3/ when faces management branch will be fixed, I would to create a new > >>> > >>> one > >>> > >>> > dedicated to 6.0.0 release and let's master for the 5.x series, as > >>> > certainly a new 5.9.0 will be published few weeks ago 5.8.0, as > >>> > maintenance, until 6.0.0 code will be stabilized. > >>> > > >>> > 4/ The clone tool : this one still not very unusable for 5.8.0. I > >>> > >>> recommend > >>> > >>> > to disable this option in editor for the moment, until zoom support > >>> > and > >>> > brushes size preview over canvas is not implemented. > >>> > > >>> > 5/ I do know yet when 5.8.0 will be released, as i don't receive a new > >>> > response from Nicolas to create the whole tarball. Also Maik, want > >>> > certainly receive more feedback about Mysql support as i can see > >>> > >>> recently > >>> > >>> > in bugzilla. > >>> > > >>> > Best > >>> > > >>> > Gilles Caulier
-- Gruß Maik |
|
|
Hum all compile fine here with OPenCV 3.4. Which version do you use ? Gilles 2018-01-07 19:55 GMT+01:00 Maik Qualmann <[hidden email]>:
|
|
|
There is nothing more in error trace, as a pure method is not implemented in EigenFaceRecognizer ? Gilles 2018-01-07 20:02 GMT+01:00 Gilles Caulier <[hidden email]>:
|
|
|
In reply to this post by Gilles Caulier-4
OpenCV-3.3.1
Maik Am Sonntag, 7. Januar 2018, 20:02:39 CET schrieb Gilles Caulier: > Hum all compile fine here with OPenCV 3.4. Which version do you use ? > > Gilles > > 2018-01-07 19:55 GMT+01:00 Maik Qualmann <[hidden email]>: > > Compile error: > > > > > > > > /home/maik/Devel/digikam-software-compilation/core/ > > libs/facesengine/recognition-opencv-eigenfaces/facerec_eigenborrowed.cpp: > > In static member function ‘static cv::P > > tr<Digikam::EigenFaceRecognizer> Digikam::EigenFaceRecognizer:: > > create(double)’: > > /home/maik/Devel/digikam-software-compilation/core/ > > libs/facesengine/recognition-opencv-eigenfaces/facerec_ > > eigenborrowed.cpp:216:70: error: invalid new-expression of > > abstract class type ‘Digikam::EigenFaceRecognizer’ > > > > EigenFaceRecognizer* const fr = new EigenFaceRecognizer(threshold); > > > > ^ > > > > Maik > > > > Am Samstag, 6. Januar 2018, 17:53:30 CET schrieb Gilles Caulier: > > > The main integration and improvements of new face recognition algorithms > > > > in > > > > > 6.0.0 branch is now complete and ready to use. > > > > > > > > > > > > The default algorithm is DNN (deep neural network => deep learning => > > > > IA). > > > > > In face scan dialog, in advanced tab, we have a new option where > > > > > > recognition algorithm can be selected. > > > > > > > > > > > > - lpbh : older one present on 5.x based on OpenCV ( > > > > > > https://docs.opencv.org/2.4/modules/contrib/doc/facerec/ > > > > facerec_tutorial.htm > > > > > l#local-binary-patterns-histograms ) > > > > > > - fisherfaces : OpenCV based ( > > > > > > https://docs.opencv.org/2.4/modules/contrib/doc/facerec/ > > > > facerec_tutorial.htm > > > > > l#fisherfaces ) > > > > > > - eigenfaces : OpenCV based ( > > > > > > https://docs.opencv.org/2.4/modules/contrib/doc/facerec/ > > > > facerec_tutorial.htm > > > > > l#eigenfaces ) > > > > > > - dnn : dlib based ( > > > > > > http://blog.dlib.net/2017/02/high-quality-face-recognition-> > > > with-deep.html) > > > > > I'm not yet satisfied by DNN. I'm not started to compare the algorithm > > > > with > > > > > a clean database. In all cases, we must detect all faces, not tagged. > > > > After > > > > > that, we need to train the faces with names, else the recognition will > > > > not > > > > > work. As i know, at least with lpbh algorithm, we need at least 6 images > > > > > > tagged with same name to see a recognition result while scanning. To > > > > have a > > > > > clean compare result, for each algorithm to use, we need a clear > > > > database. > > > > > The student working on this project while summer 2017 report to me that > > > > in > > > > > all cases the best algorithms are : > > > > > > > > > > > > 1 DNN. > > > > > > 2 eigenfaces. > > > > > > 3 fisherfaces. > > > > > > 4 lpbh. > > > > > > > > > > > > The goal is to use the best one at end. > > > > > > > > > > > > Gilles Caulier > > > > > > 2018-01-04 13:18 GMT+01:00 Gilles Caulier <[hidden email]>: > > > > The 6.0.0 git branch is created in remote : > > > > > > > > > > > > > > > > [gilles@localhost digikam.master]$ git branch -a > > > > > > > > * master > > > > > > > > > > > > > > > > remotes/origin/HEAD -> origin/master > > > > > > > > remotes/origin/development/0.10 > > > > > > > > remotes/origin/development/0.8 > > > > > > > > remotes/origin/development/0.8.x > > > > > > > > remotes/origin/development/0.9.x > > > > > > > > remotes/origin/development/1.2 > > > > > > > > remotes/origin/development/1.3 > > > > > > > > remotes/origin/development/3.0.0 > > > > > > > > remotes/origin/development/6.0.0 < > > > > > > > > > > > > > > > > ============================= > > > > > > > > > > > > > > > > remotes/origin/development/balooport > > > > > > > > remotes/origin/development/bug309058 > > > > > > > > remotes/origin/development/cmakecleanup > > > > > > > > remotes/origin/development/externaltools > > > > > > > > remotes/origin/development/frameworks > > > > > > > > remotes/origin/development/garbagecollection > > > > > > > > remotes/origin/development/hupnp > > > > > > > > remotes/origin/development/imagemetadatacache > > > > > > > > remotes/origin/development/kde3 > > > > > > > > remotes/origin/development/photolayoutseditor > > > > > > > > remotes/origin/development/sql2.0 > > > > > > > > remotes/origin/development/usbimport > > > > > > > > remotes/origin/gsoc11-clone > > > > > > > > remotes/origin/gsoc11-presentation > > > > > > > > remotes/origin/gsoc13-editorcanvas > > > > > > > > remotes/origin/gsoc13-imgqsort > > > > > > > > remotes/origin/gsoc13-localadj > > > > > > > > remotes/origin/gsoc13-tagmngr > > > > > > > > remotes/origin/gsoc13-thumbbar-mv > > > > > > > > remotes/origin/gsoc13-videow > > > > > > > > remotes/origin/gsoc14-greyc > > > > > > > > remotes/origin/gsoc14-kipibqm > > > > > > > > remotes/origin/gsoc14-qaccess > > > > > > > > remotes/origin/gsoc15-advmetadata > > > > > > > > remotes/origin/gsoc15-kio-drop > > > > > > > > remotes/origin/gsoc16-mysql > > > > > > > > remotes/origin/gsoc16-red-eye > > > > > > > > remotes/origin/gsoc17-db-sep > > > > > > > > remotes/origin/gsoc17-dlna > > > > > > > > remotes/origin/gsoc17-face-mngmnt > > > > > > > > remotes/origin/gsoc17-healing > > > > > > > > remotes/origin/master > > > > > > > > > > > > > > > > This one is based on gsoc17-face-mngmnt sync with master today > > > > morning. > > > > > > > > > > > > > > > > The remotes/origin/development/6.0.0 must be used to put next 6.0.0 > > > > new > > > > > > code and must be sync with master until 5.x serie will be completed. > > > > > > > > > > > > > > > > Best > > > > > > > > > > > > > > > > Gilles Caulier > > > > > > > > 2018-01-04 12:24 GMT+01:00 Gilles Caulier <[hidden email]>: > > > >> The branch gsoc17-face-mngmnt is now sync with master and compile and > > > >> > > > >> link properly. > > > >> > > > >> > > > >> > > > >> I will create a new 6.0.0 branch based on the gsoc17-face-mngmnt > > > > branch > > > > > >> for future main development. > > > >> > > > >> > > > >> > > > >> Gilles Caulier > > > >> > > > >> 2018-01-03 19:57 GMT+01:00 Maik Qualmann <[hidden email]>: > > > >>> Gilles, and everyone else a Happy New Year. Yes, I'm looking forward > > > > to > > > > > >>> feedback on MySQL - but only positive. ((:-)) No, seriously, I know > > > > that > > > > > >>> there > > > >>> > > > >>> will be problems. Currently I only use MySQL for testing and it is > > > >>> > > > >>> running > > > >>> > > > >>> really well here. It is important that we write in the release note > > > > that > > > > > >>> the > > > >>> > > > >>> core DB with the migration tool should be copied into a new DB and > > > >>> > > > >>> should be > > > >>> > > > >>> started with a new created thumbnail and face DB. > > > >>> > > > >>> > > > >>> > > > >>> Maik > > > >>> > > > >>> Am Mittwoch, 3. Januar 2018, 18:10:50 CET schrieb Gilles Caulier: > > > >>> > Hi all, > > > >>> > > > > >>> > > > > >>> > > > > >>> > In first, i hope an happy new year to all. > > > >>> > > > > >>> > > > > >>> > > > > >>> > 1/ I currently revival the face management git branch written > > > >>> > while > > > >>> > > > >>> summer > > > >>> > > > >>> > 2017 by the student working on AI engine to recognize faces. The > > > > code > > > > > >>> is > > > >>> > > > >>> > sync with master, but do not compile yet. I will fix it. > > > >>> > > > > >>> > > > > >>> > > > > >>> > 2/ Important point : this code do not compile with OpenCV 2.x. It > > > >>> > > > >>> require > > > >>> > > > >>> > OpenCV 3.1. This want mean that digiKam 6.0.0 will drop OpenCV 2.x > > > >>> > > > >>> support > > > >>> > > > >>> > definitively. > > > >>> > > > > >>> > > > > >>> > > > > >>> > 3/ when faces management branch will be fixed, I would to create a > > > > new > > > > > >>> one > > > >>> > > > >>> > dedicated to 6.0.0 release and let's master for the 5.x series, as > > > >>> > > > > >>> > certainly a new 5.9.0 will be published few weeks ago 5.8.0, as > > > >>> > > > > >>> > maintenance, until 6.0.0 code will be stabilized. > > > >>> > > > > >>> > > > > >>> > > > > >>> > 4/ The clone tool : this one still not very unusable for 5.8.0. I > > > >>> > > > >>> recommend > > > >>> > > > >>> > to disable this option in editor for the moment, until zoom > > > >>> > support > > > >>> > > > > >>> > and > > > >>> > > > > >>> > brushes size preview over canvas is not implemented. > > > >>> > > > > >>> > > > > >>> > > > > >>> > 5/ I do know yet when 5.8.0 will be released, as i don't receive a > > > > new > > > > > >>> > response from Nicolas to create the whole tarball. Also Maik, want > > > >>> > > > > >>> > certainly receive more feedback about Mysql support as i can see > > > >>> > > > >>> recently > > > >>> > > > >>> > in bugzilla. > > > >>> > > > > >>> > > > > >>> > > > > >>> > Best > > > >>> > > > > >>> > > > > >>> > > > > >>> > Gilles Caulier > > > > -- > > > > Gruß Maik -- Gruß Maik |
|
|
In reply to this post by Gilles Caulier-4
I'm just compiling master again. I'll take another look at the days, right now
I'm back to MySQL... Maik Am Sonntag, 7. Januar 2018, 20:05:50 CET schrieb Gilles Caulier: > There is nothing more in error trace, as a pure method is not implemented > in EigenFaceRecognizer ? > > Gilles > > 2018-01-07 20:02 GMT+01:00 Gilles Caulier <[hidden email]>: > > Hum all compile fine here with OPenCV 3.4. Which version do you use ? > > > > Gilles > > > > 2018-01-07 19:55 GMT+01:00 Maik Qualmann <[hidden email]>: > >> Compile error: > >> > >> > >> > >> /home/maik/Devel/digikam-software-compilation/core/libs/ > >> facesengine/recognition-opencv-eigenfaces/facerec_eigenborrowed.cpp: In > >> static member function ‘static cv::P > >> tr<Digikam::EigenFaceRecognizer> Digikam::EigenFaceRecognizer:: > >> create(double)’: > >> /home/maik/Devel/digikam-software-compilation/core/libs/ > >> facesengine/recognition-opencv-eigenfaces/facerec_eigenborro > >> wed.cpp:216:70: error: invalid new-expression of > >> abstract class type ‘Digikam::EigenFaceRecognizer’ > >> > >> EigenFaceRecognizer* const fr = new EigenFaceRecognizer(threshold); > >> > >> ^ > >> > >> Maik > >> > >> Am Samstag, 6. Januar 2018, 17:53:30 CET schrieb Gilles Caulier: > >> > The main integration and improvements of new face recognition > >> > >> algorithms in > >> > >> > 6.0.0 branch is now complete and ready to use. > >> > > >> > > >> > > >> > The default algorithm is DNN (deep neural network => deep learning => > >> > >> IA). > >> > >> > In face scan dialog, in advanced tab, we have a new option where > >> > > >> > recognition algorithm can be selected. > >> > > >> > > >> > > >> > - lpbh : older one present on 5.x based on OpenCV ( > >> > > >> > https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face > >> > >> rec_tutorial.htm > >> > >> > l#local-binary-patterns-histograms ) > >> > > >> > - fisherfaces : OpenCV based ( > >> > > >> > https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face > >> > >> rec_tutorial.htm > >> > >> > l#fisherfaces ) > >> > > >> > - eigenfaces : OpenCV based ( > >> > > >> > https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face > >> > >> rec_tutorial.htm > >> > >> > l#eigenfaces ) > >> > > >> > - dnn : dlib based ( > >> > > >> > http://blog.dlib.net/2017/02/high-quality-face-recognition-w > >> > >> ith-deep.html) > >> > >> > I'm not yet satisfied by DNN. I'm not started to compare the algorithm > >> > >> with > >> > >> > a clean database. In all cases, we must detect all faces, not tagged. > >> > >> After > >> > >> > that, we need to train the faces with names, else the recognition will > >> > >> not > >> > >> > work. As i know, at least with lpbh algorithm, we need at least 6 > >> > images > >> > > >> > tagged with same name to see a recognition result while scanning. To > >> > >> have a > >> > >> > clean compare result, for each algorithm to use, we need a clear > >> > >> database. > >> > >> > The student working on this project while summer 2017 report to me that > >> > >> in > >> > >> > all cases the best algorithms are : > >> > > >> > > >> > > >> > 1 DNN. > >> > > >> > 2 eigenfaces. > >> > > >> > 3 fisherfaces. > >> > > >> > 4 lpbh. > >> > > >> > > >> > > >> > The goal is to use the best one at end. > >> > > >> > > >> > > >> > Gilles Caulier > >> > > >> > 2018-01-04 13:18 GMT+01:00 Gilles Caulier <[hidden email]>: > >> > > The 6.0.0 git branch is created in remote : > >> > > > >> > > > >> > > > >> > > [gilles@localhost digikam.master]$ git branch -a > >> > > > >> > > * master > >> > > > >> > > > >> > > > >> > > remotes/origin/HEAD -> origin/master > >> > > > >> > > remotes/origin/development/0.10 > >> > > > >> > > remotes/origin/development/0.8 > >> > > > >> > > remotes/origin/development/0.8.x > >> > > > >> > > remotes/origin/development/0.9.x > >> > > > >> > > remotes/origin/development/1.2 > >> > > > >> > > remotes/origin/development/1.3 > >> > > > >> > > remotes/origin/development/3.0.0 > >> > > > >> > > remotes/origin/development/6.0.0 < > >> > > > >> > > > >> > > > >> > > ============================= > >> > > > >> > > > >> > > > >> > > remotes/origin/development/balooport > >> > > > >> > > remotes/origin/development/bug309058 > >> > > > >> > > remotes/origin/development/cmakecleanup > >> > > > >> > > remotes/origin/development/externaltools > >> > > > >> > > remotes/origin/development/frameworks > >> > > > >> > > remotes/origin/development/garbagecollection > >> > > > >> > > remotes/origin/development/hupnp > >> > > > >> > > remotes/origin/development/imagemetadatacache > >> > > > >> > > remotes/origin/development/kde3 > >> > > > >> > > remotes/origin/development/photolayoutseditor > >> > > > >> > > remotes/origin/development/sql2.0 > >> > > > >> > > remotes/origin/development/usbimport > >> > > > >> > > remotes/origin/gsoc11-clone > >> > > > >> > > remotes/origin/gsoc11-presentation > >> > > > >> > > remotes/origin/gsoc13-editorcanvas > >> > > > >> > > remotes/origin/gsoc13-imgqsort > >> > > > >> > > remotes/origin/gsoc13-localadj > >> > > > >> > > remotes/origin/gsoc13-tagmngr > >> > > > >> > > remotes/origin/gsoc13-thumbbar-mv > >> > > > >> > > remotes/origin/gsoc13-videow > >> > > > >> > > remotes/origin/gsoc14-greyc > >> > > > >> > > remotes/origin/gsoc14-kipibqm > >> > > > >> > > remotes/origin/gsoc14-qaccess > >> > > > >> > > remotes/origin/gsoc15-advmetadata > >> > > > >> > > remotes/origin/gsoc15-kio-drop > >> > > > >> > > remotes/origin/gsoc16-mysql > >> > > > >> > > remotes/origin/gsoc16-red-eye > >> > > > >> > > remotes/origin/gsoc17-db-sep > >> > > > >> > > remotes/origin/gsoc17-dlna > >> > > > >> > > remotes/origin/gsoc17-face-mngmnt > >> > > > >> > > remotes/origin/gsoc17-healing > >> > > > >> > > remotes/origin/master > >> > > > >> > > > >> > > > >> > > This one is based on gsoc17-face-mngmnt sync with master today > >> > >> morning. > >> > >> > > The remotes/origin/development/6.0.0 must be used to put next 6.0.0 > >> > >> new > >> > >> > > code and must be sync with master until 5.x serie will be completed. > >> > > > >> > > > >> > > > >> > > Best > >> > > > >> > > > >> > > > >> > > Gilles Caulier > >> > > > >> > > 2018-01-04 12:24 GMT+01:00 Gilles Caulier <[hidden email]>: > >> > >> The branch gsoc17-face-mngmnt is now sync with master and compile > >> > >> and > >> > >> > >> > >> link properly. > >> > >> > >> > >> > >> > >> > >> > >> I will create a new 6.0.0 branch based on the gsoc17-face-mngmnt > >> > >> branch > >> > >> > >> for future main development. > >> > >> > >> > >> > >> > >> > >> > >> Gilles Caulier > >> > >> > >> > >> 2018-01-03 19:57 GMT+01:00 Maik Qualmann <[hidden email]>: > >> > >>> Gilles, and everyone else a Happy New Year. Yes, I'm looking > >> > >> forward to > >> > >> > >>> feedback on MySQL - but only positive. ((:-)) No, seriously, I know > >> > >> that > >> > >> > >>> there > >> > >>> > >> > >>> will be problems. Currently I only use MySQL for testing and it is > >> > >>> > >> > >>> running > >> > >>> > >> > >>> really well here. It is important that we write in the release note > >> > >> that > >> > >> > >>> the > >> > >>> > >> > >>> core DB with the migration tool should be copied into a new DB and > >> > >>> > >> > >>> should be > >> > >>> > >> > >>> started with a new created thumbnail and face DB. > >> > >>> > >> > >>> > >> > >>> > >> > >>> Maik > >> > >>> > >> > >>> Am Mittwoch, 3. Januar 2018, 18:10:50 CET schrieb Gilles Caulier: > >> > >>> > Hi all, > >> > >>> > > >> > >>> > > >> > >>> > > >> > >>> > In first, i hope an happy new year to all. > >> > >>> > > >> > >>> > > >> > >>> > > >> > >>> > 1/ I currently revival the face management git branch written > >> > >> while > >> > >> > >>> summer > >> > >>> > >> > >>> > 2017 by the student working on AI engine to recognize faces. The > >> > >> code > >> > >> > >>> is > >> > >>> > >> > >>> > sync with master, but do not compile yet. I will fix it. > >> > >>> > > >> > >>> > > >> > >>> > > >> > >>> > 2/ Important point : this code do not compile with OpenCV 2.x. It > >> > >>> > >> > >>> require > >> > >>> > >> > >>> > OpenCV 3.1. This want mean that digiKam 6.0.0 will drop OpenCV > >> > >>> > 2.x > >> > >>> > >> > >>> support > >> > >>> > >> > >>> > definitively. > >> > >>> > > >> > >>> > > >> > >>> > > >> > >>> > 3/ when faces management branch will be fixed, I would to create > >> > >> a new > >> > >> > >>> one > >> > >>> > >> > >>> > dedicated to 6.0.0 release and let's master for the 5.x series, > >> > >>> > as > >> > >>> > > >> > >>> > certainly a new 5.9.0 will be published few weeks ago 5.8.0, as > >> > >>> > > >> > >>> > maintenance, until 6.0.0 code will be stabilized. > >> > >>> > > >> > >>> > > >> > >>> > > >> > >>> > 4/ The clone tool : this one still not very unusable for 5.8.0. I > >> > >>> > >> > >>> recommend > >> > >>> > >> > >>> > to disable this option in editor for the moment, until zoom > >> > >> support > >> > >> > >>> > and > >> > >>> > > >> > >>> > brushes size preview over canvas is not implemented. > >> > >>> > > >> > >>> > > >> > >>> > > >> > >>> > 5/ I do know yet when 5.8.0 will be released, as i don't receive > >> > >> a new > >> > >> > >>> > response from Nicolas to create the whole tarball. Also Maik, > >> > >>> > want > >> > >>> > > >> > >>> > certainly receive more feedback about Mysql support as i can see > >> > >>> > >> > >>> recently > >> > >>> > >> > >>> > in bugzilla. > >> > >>> > > >> > >>> > > >> > >>> > > >> > >>> > Best > >> > >>> > > >> > >>> > > >> > >>> > > >> > >>> > Gilles Caulier > >> > >> -- > >> > >> Gruß Maik |
|
|
In reply to this post by Gilles Caulier-4
Gilles, are any additional developer packages needed? I have many notes here and then errors in:
core/libs/facesengine/dnnface/dnn_base/serialize.h
That's just a small excerpt:
/home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dnnface/nn/../matrix/../dnn_base/serialize.h:443:21: note: candidate: void serialize(const short u In file included from /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/facedb/facedb.cpp:28:0:
Maik Am Sonntag, 7. Januar 2018, 20:05:50 CET schrieb Gilles Caulier: > There is nothing more in error trace, as a pure method is not implemented > in EigenFaceRecognizer ? > > Gilles > > 2018-01-07 20:02 GMT+01:00 Gilles Caulier <[hidden email]>: > > Hum all compile fine here with OPenCV 3.4. Which version do you use ? > > > > Gilles > > > > 2018-01-07 19:55 GMT+01:00 Maik Qualmann <[hidden email]>: > >> Compile error: > >> > >> > >> > >> /home/maik/Devel/digikam-software-compilation/core/libs/ > >> facesengine/recognition-opencv-eigenfaces/facerec_eigenborrowed.cpp: In > >> static member function ‘static cv::P > >> tr<Digikam::EigenFaceRecognizer> Digikam::EigenFaceRecognizer:: > >> create(double)’: > >> /home/maik/Devel/digikam-software-compilation/core/libs/ > >> facesengine/recognition-opencv-eigenfaces/facerec_eigenborro > >> wed.cpp:216:70: error: invalid new-expression of > >> abstract class type ‘Digikam::EigenFaceRecognizer’ > >> > >> EigenFaceRecognizer* const fr = new EigenFaceRecognizer(threshold); > >> > >> ^ > >> > >> Maik > >> > >> Am Samstag, 6. Januar 2018, 17:53:30 CET schrieb Gilles Caulier: > >> > The main integration and improvements of new face recognition > >> > >> algorithms in > >> > >> > 6.0.0 branch is now complete and ready to use. > >> > > >> > > >> > > >> > The default algorithm is DNN (deep neural network => deep learning => > >> > >> IA). > >> > >> > In face scan dialog, in advanced tab, we have a new option where > >> > > >> > recognition algorithm can be selected. > >> > > >> > > >> > > >> > - lpbh : older one present on 5.x based on OpenCV ( > >> > > >> > https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face > >> > >> rec_tutorial.htm > >> > >> > l#local-binary-patterns-histograms ) > >> > > >> > - fisherfaces : OpenCV based ( > >> > > >> > https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face > >> > >> rec_tutorial.htm > >> > >> > l#fisherfaces ) > >> > > >> > - eigenfaces : OpenCV based ( > >> > > >> > https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face > >> > >> rec_tutorial.htm > >> > >> > l#eigenfaces ) > >> > > >> > - dnn : dlib based ( > >> > > >> > http://blog.dlib.net/2017/02/high-quality-face-recognition-w > >> > >> ith-deep.html) > >> > >> > I'm not yet satisfied by DNN. I'm not started to compare the algorithm > >> > >> with > >> > >> > a clean database. In all cases, we must detect all faces, not tagged. > >> > >> After > >> > >> > that, we need to train the faces with names, else the recognition will > >> > >> not > >> > >> > work. As i know, at least with lpbh algorithm, we need at least 6 > >> > images > >> > > >> > tagged with same name to see a recognition result while scanning. To > >> > >> have a > >> > >> > clean compare result, for each algorithm to use, we need a clear > >> > >> database. > >> > >> > The student working on this project while summer 2017 report to me that > >> > >> in > >> > >> > all cases the best algorithms are : > >> > > >> > > >> > > >> > 1 DNN. > >> > > >> > 2 eigenfaces. > >> > > >> > 3 fisherfaces. > >> > > >> > 4 lpbh. > >> > > >> > > >> > > >> > The goal is to use the best one at end. > >> > > >> > > >> > > >> > Gilles Caulier > >> > > >> > 2018-01-04 13:18 GMT+01:00 Gilles Caulier <[hidden email]>: > >> > > The 6.0.0 git branch is created in remote : > >> > > > >> > > > >> > > > >> > > [gilles@localhost digikam.master]$ git branch -a > >> > > > >> > > * master > >> > > > >> > > > >> > > > >> > > remotes/origin/HEAD -> origin/master > >> > > > >> > > remotes/origin/development/0.10 > >> > > > >> > > remotes/origin/development/0.8 > >> > > > >> > > remotes/origin/development/0.8.x > >> > > > >> > > remotes/origin/development/0.9.x > >> > > > >> > > remotes/origin/development/1.2 > >> > > > >> > > remotes/origin/development/1.3 > >> > > > >> > > remotes/origin/development/3.0.0 > >> > > > >> > > remotes/origin/development/6.0.0 < > >> > > > >> > > > >> > > > >> > > ============================= > >> > > > >> > > > >> > > > >> > > remotes/origin/development/balooport > >> > > > >> > > remotes/origin/development/bug309058 > >> > > > >> > > remotes/origin/development/cmakecleanup > >> > > > >> > > remotes/origin/development/externaltools > >> > > > >> > > remotes/origin/development/frameworks > >> > > > >> > > remotes/origin/development/garbagecollection > >> > > > >> > > remotes/origin/development/hupnp > >> > > > >> > > remotes/origin/development/imagemetadatacache > >> > > > >> > > remotes/origin/development/kde3 > >> > > > >> > > remotes/origin/development/photolayoutseditor > >> > > > >> > > remotes/origin/development/sql2.0 > >> > > > >> > > remotes/origin/development/usbimport > >> > > > >> > > remotes/origin/gsoc11-clone > >> > > > >> > > remotes/origin/gsoc11-presentation > >> > > > >> > > remotes/origin/gsoc13-editorcanvas > >> > > > >> > > remotes/origin/gsoc13-imgqsort > >> > > > >> > > remotes/origin/gsoc13-localadj > >> > > > >> > > remotes/origin/gsoc13-tagmngr > >> > > > >> > > remotes/origin/gsoc13-thumbbar-mv > >> > > > >> > > remotes/origin/gsoc13-videow > >> > > > >> > > remotes/origin/gsoc14-greyc > >> > > > >> > > remotes/origin/gsoc14-kipibqm > >> > > > >> > > remotes/origin/gsoc14-qaccess > >> > > > >> > > remotes/origin/gsoc15-advmetadata > >> > > > >> > > remotes/origin/gsoc15-kio-drop > >> > > > >> > > remotes/origin/gsoc16-mysql > >> > > > >> > > remotes/origin/gsoc16-red-eye > >> > > > >> > > remotes/origin/gsoc17-db-sep > >> > > > >> > > remotes/origin/gsoc17-dlna > >> > > > >> > > remotes/origin/gsoc17-face-mngmnt > >> > > > >> > > remotes/origin/gsoc17-healing > >> > > > >> > > remotes/origin/master > >> > > > >> > > > >> > > > >> > > This one is based on gsoc17-face-mngmnt sync with master today > >> > >> morning. > >> > >> > > The remotes/origin/development/6.0.0 must be used to put next 6.0.0 > >> > >> new > >> > >> > > code and must be sync with master until 5.x serie will be completed. > >> > > > >> > > > >> > > > >> > > Best > >> > > > >> > > > >> > > > >> > > Gilles Caulier > >> > > > >> > > 2018-01-04 12:24 GMT+01:00 Gilles Caulier <[hidden email]>: > >> > >> The branch gsoc17-face-mngmnt is now sync with master and compile > >> > >> and > >> > >> > >> > >> link properly. > >> > >> > >> > >> > >> > >> > >> > >> I will create a new 6.0.0 branch based on the gsoc17-face-mngmnt > >> > >> branch > >> > >> > >> for future main development. > >> > >> > >> > >> > >> > >> > >> > >> Gilles Caulier > >> > >> > >> > >> 2018-01-03 19:57 GMT+01:00 Maik Qualmann <[hidden email]>: > >> > >>> Gilles, and everyone else a Happy New Year. Yes, I'm looking > >> > >> forward to > >> > >> > >>> feedback on MySQL - but only positive. ((:-)) No, seriously, I know > >> > >> that > >> > >> > >>> there > >> > >>> > >> > >>> will be problems. Currently I only use MySQL for testing and it is > >> > >>> > >> > >>> running > >> > >>> > >> > >>> really well here. It is important that we write in the release note > >> > >> that > >> > >> > >>> the > >> > >>> > >> > >>> core DB with the migration tool should be copied into a new DB and > >> > >>> > >> > >>> should be > >> > >>> > >> > >>> started with a new created thumbnail and face DB. > >> > >>> > >> > >>> > >> > >>> > >> > >>> Maik > >> > >>> > >> > >>> Am Mittwoch, 3. Januar 2018, 18:10:50 CET schrieb Gilles Caulier: > >> > >>> > Hi all, > >> > >>> > > >> > >>> > > >> > >>> > > >> > >>> > In first, i hope an happy new year to all. > >> > >>> > > >> > >>> > > >> > >>> > > >> > >>> > 1/ I currently revival the face management git branch written > >> > >> while > >> > >> > >>> summer > >> > >>> > >> > >>> > 2017 by the student working on AI engine to recognize faces. The > >> > >> code > >> > >> > >>> is > >> > >>> > >> > >>> > sync with master, but do not compile yet. I will fix it. > >> > >>> > > >> > >>> > > >> > >>> > > >> > >>> > 2/ Important point : this code do not compile with OpenCV 2.x. It > >> > >>> > >> > >>> require > >> > >>> > >> > >>> > OpenCV 3.1. This want mean that digiKam 6.0.0 will drop OpenCV > >> > >>> > 2.x > >> > >>> > >> > >>> support > >> > >>> > >> > >>> > definitively. > >> > >>> > > >> > >>> > > >> > >>> > > >> > >>> > 3/ when faces management branch will be fixed, I would to create > >> > >> a new > >> > >> > >>> one > >> > >>> > >> > >>> > dedicated to 6.0.0 release and let's master for the 5.x series, > >> > >>> > as > >> > >>> > > >> > >>> > certainly a new 5.9.0 will be published few weeks ago 5.8.0, as > >> > >>> > > >> > >>> > maintenance, until 6.0.0 code will be stabilized. > >> > >>> > > >> > >>> > > >> > >>> > > >> > >>> > 4/ The clone tool : this one still not very unusable for 5.8.0. I > >> > >>> > >> > >>> recommend > >> > >>> > >> > >>> > to disable this option in editor for the moment, until zoom > >> > >> support > >> > >> > >>> > and > >> > >>> > > >> > >>> > brushes size preview over canvas is not implemented. > >> > >>> > > >> > >>> > > >> > >>> > > >> > >>> > 5/ I do know yet when 5.8.0 will be released, as i don't receive > >> > >> a new > >> > >> > >>> > response from Nicolas to create the whole tarball. Also Maik, > >> > >>> > want > >> > >>> > > >> > >>> > certainly receive more feedback about Mysql support as i can see > >> > >>> > >> > >>> recently > >> > >>> > >> > >>> > in bugzilla. > >> > >>> > > >> > >>> > > >> > >>> > > >> > >>> > Best > >> > >>> > > >> > >>> > > >> > >>> > > >> > >>> > Gilles Caulier > >> > >> -- > >> > >> Gruß Maik |
Re: preparing 6.0.0 code...
|
|
In reply to this post by Gilles Caulier-4
Hi Gilles and all,
hope all is going good! May i suggest to redraw the kipi logo inside the settings dialog? (Settings->Configure Digikam->Plugins) it isn't very good looking, I mean not the logo itself (i remember the time in which we drew it :D), but the icon/png/resource or what is to be included there. My best, Angelo |
|
|
In reply to this post by Maik Qualmann
Hi Maik, none. all compile fine here without extra package. Which GCC version do you use exactly ? Gilles 2018-01-11 12:41 GMT+01:00 Maik Qualmann <[hidden email]>:
|
|
|
In reply to this post by Bugzilla from anaselli@linux.it
Hi Angelo, Gilles 2018-01-11 14:30 GMT+01:00 Angelo Naselli <[hidden email]>: Hi Gilles and all, |
Re: preparing 6.0.0 code...
|
|
Il 11/01/2018 16:17, Gilles Caulier ha scritto:
> Hi Angelo, Hey! > yes, why not to redraw this icon. The file is available here : > > https://cgit.kde.org/libkipi.git/tree/pics Well it isn't the icon di per se, but the way it is shown that is pretty ugly. I can provide a screenshot if you need to..., Maybe load it as svg? I mean we show beautiful picture preview, and our logo seems out of resolution... |
|
|
In reply to this post by Gilles Caulier-4
I have the same problem, the first error I get is:
In file included from /media/nunkystorage/4_Simon/Programme/Linux/source/digikam/develsoftwarecomp/core/libs/facesengine/facedb/facedb.cpp:28:0: /media/nunkystorage/4_Simon/Programme/Linux/source/digikam/develsoftwarecomp/core/libs/facesengine/dnnface/nn/layers.h: In function 'void serialize(const con_<_num_filters, _nr, _nc, _stride_y, _stride_x, _padding_y, _padding_x>&, std::ostream&)': /media/nunkystorage/4_Simon/Programme/Linux/source/digikam/develsoftwarecomp/core/libs/facesengine/dnnface/nn/layers.h:210:39: error: no matching function for call to 'serialize(const resizable_tensor&, std::ostream&)' serialize(item.params, out); ^ Then lots of candidates for serialize functions, but there is no serialize function defined for "const resizable_tensor&" as argument. So I am a bit confused why it compiles for you at all. Just in case, I am on gcc version 7.2. On 11/01/18 16:14, Gilles Caulier
wrote:
|
|
|
Yes, here also GCC 7.2.1. At the moment I am lost here too...
Maik Am Donnerstag, 11. Januar 2018, 17:29:34 CET schrieb Simon Frei: > I have the same problem, the first error I get is: > > In file included from > /media/nunkystorage/4_Simon/Programme/Linux/source/digikam/develsoftwarecomp > /core/libs/facesengine/facedb/facedb.cpp:28:0: > /media/nunkystorage/4_Simon/Programme/Linux/source/digikam/develsoftwarecom > p/core/libs/facesengine/dnnface/nn/layers.h: In function 'void > serialize(const con_<_num_filters, _nr, _nc, > _stride_y, _stride_x, _padding_y, _padding_x>&, std::ostream&)': > /media/nunkystorage/4_Simon/Programme/Linux/source/digikam/develsoftwarecomp > /core/libs/facesengine/dnnface/nn/layers.h:210:39: error: no matching > function for call to 'serialize(const > resizable_tensor&, std::ostream&)' > serialize(item.params, out); > ^ > Then lots of candidates for serialize functions, but there is no > serialize function defined for "const resizable_tensor&" as argument. So > I am a bit confused why it compiles for you at all. Just in case, I am > on gcc version 7.2. > > On 11/01/18 16:14, Gilles Caulier wrote: > > Hi Maik, > > > > none. all compile fine here without extra package. > > > > Which GCC version do you use exactly ? > > > > Gilles > > > > 2018-01-11 12:41 GMT+01:00 Maik Qualmann <[hidden email] > > > > <mailto:[hidden email]>>: > > Gilles, are any additional developer packages needed? I have many > > notes here and then errors in: > > > > > > > > core/libs/facesengine/dnnface/dnn_base/serialize.h > > > > > > > > That's just a small excerpt: > > > > > > > > /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dn > > nface/nn/../matrix/../dnn_base/serialize.h:443:21:note: candidate: > > void serialize(const short u > > nsigned int&, std::ostream&) > > inline void serialize (const T& item, std::ostream& out) \ > > ^ > > /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dn > > nface/nn/../matrix/../dnn_base/serialize.h:504:5:note: in expansion of > > macro ‘USE_DEFAULT_INT_S > > ERIALIZATION_FOR’ > > USE_DEFAULT_INT_SERIALIZATION_FOR(unsigned short) > > ^ > > /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dn > > nface/nn/../matrix/../dnn_base/serialize.h:443:21:note: no known > > conversion for argument 1 fr > > om ‘const alias_tensor’ to ‘const short unsigned int&’ > > inline void serialize (const T& item, std::ostream& out) \ > > > > In file included from > > /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/fa > > cedb/facedb.cpp:28:0: > > > > /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dn > > nface/nn/layers.h:1162:39:error: no matching function for call to > > ‘serialize(const alias_tensor > > &, std::ostream&)’ > > serialize(item.biases, out); > > > > /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dn > > nface/nn/../matrix/../dnn_base/serialize.h:1596:54:note: template > > argument deduction/substitu > > tion failed: > > /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dn > > nface/nn/../matrix/../dnn_base/serialize.h: In substitution of > > ‘template<class T> typename enabl > > e_if<is_protocol_buffer<T> >::type serialize(const T&, > > std::ostream&) [with T = alias_tensor]’: > > /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dn > > nface/nn/layers.h:1162:39: required from here > > /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dn > > nface/nn/../matrix/../dnn_base/serialize.h:1596:54:error: no type > > named ‘type’ in ‘struct enabl > > e_if<is_protocol_buffer<alias_tensor, void>, void>’ > > > > > > > > Maik > > > > Am Sonntag, 7. Januar 2018, 20:05:50 CET schrieb Gilles Caulier: > > > There is nothing more in error trace, as a pure method is not > > > > implemented > > > > > in EigenFaceRecognizer ? > > > > > > > > > > > > Gilles > > > > > > > > > > > > 2018-01-07 20:02 GMT+01:00 Gilles Caulier > > > > <[hidden email] <mailto:[hidden email]>>: > > > > Hum all compile fine here with OPenCV 3.4. Which version do > > > > you use ? > > > > > > Gilles > > > > > > > > > > > > > > > > 2018-01-07 19:55 GMT+01:00 Maik Qualmann > > > > <[hidden email] <mailto:[hidden email]>>: > > > >> Compile error: > > > >> > > > >> > > > >> > > > >> > > > >> > > > >> > > > >> > > > >> /home/maik/Devel/digikam-software-compilation/core/libs/ > > > > facesengine/recognition-opencv-eigenfaces/facerec_eigenborrowed.cpp: > > In > > > > > >> static member function ‘static cv::P > > > >> > > > >> tr<Digikam::EigenFaceRecognizer> Digikam::EigenFaceRecognizer:: > > > >> > > > >> create(double)’: > > > >> > > > >> /home/maik/Devel/digikam-software-compilation/core/libs/ > > > >> > > > >> facesengine/recognition-opencv-eigenfaces/facerec_eigenborro > > > >> > > > >> wed.cpp:216:70: error: invalid new-expression of > > > >> > > > >> abstract class type ‘Digikam::EigenFaceRecognizer’ > > > >> > > > >> > > > >> > > > >> EigenFaceRecognizer* const fr = new > > > > EigenFaceRecognizer(threshold); > > > > > >> ^ > > > >> > > > >> > > > >> > > > >> Maik > > > >> > > > >> Am Samstag, 6. Januar 2018, 17:53:30 CET schrieb Gilles Caulier: > > > >> > The main integration and improvements of new face recognition > > > >> > > > >> algorithms in > > > >> > > > >> > 6.0.0 branch is now complete and ready to use. > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > The default algorithm is DNN (deep neural network => deep > > > > learning => > > > > > >> IA). > > > >> > > > >> > In face scan dialog, in advanced tab, we have a new option > > > > where > > > > > >> > recognition algorithm can be selected. > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > - lpbh : older one present on 5.x based on OpenCV ( > > > > https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face > > <https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face> > > > > > >> rec_tutorial.htm > > > >> > > > >> > l#local-binary-patterns-histograms ) > > > >> > > > > >> > > > > >> > > > > >> > - fisherfaces : OpenCV based ( > > > > https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face > > <https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face> > > > > > >> rec_tutorial.htm > > > >> > > > >> > l#fisherfaces ) > > > >> > > > > >> > > > > >> > > > > >> > - eigenfaces : OpenCV based ( > > > > https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face > > <https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face> > > > > > >> rec_tutorial.htm > > > >> > > > >> > l#eigenfaces ) > > > >> > > > > >> > > > > >> > > > > >> > - dnn : dlib based ( > > > > http://blog.dlib.net/2017/02/high-quality-face-recognition-w > > <http://blog.dlib.net/2017/02/high-quality-face-recognition-w> > > > > > >> ith-deep.html) > > > >> > > > >> > I'm not yet satisfied by DNN. I'm not started to compare > > > > the algorithm > > > > > >> with > > > >> > > > >> > a clean database. In all cases, we must detect all faces, > > > > not tagged. > > > > > >> After > > > >> > > > >> > that, we need to train the faces with names, else the > > > > recognition will > > > > > >> not > > > >> > > > >> > work. As i know, at least with lpbh algorithm, we need at > > > > least 6 > > > > > >> > images > > > >> > > > > >> > > > > >> > > > > >> > tagged with same name to see a recognition result while > > > > scanning. To > > > > > >> have a > > > >> > > > >> > clean compare result, for each algorithm to use, we need a > > > > clear > > > > > >> database. > > > >> > > > >> > The student working on this project while summer 2017 > > > > report to me that > > > > > >> in > > > >> > > > >> > all cases the best algorithms are : > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > 1 DNN. > > > >> > > > > >> > > > > >> > > > > >> > 2 eigenfaces. > > > >> > > > > >> > > > > >> > > > > >> > 3 fisherfaces. > > > >> > > > > >> > > > > >> > > > > >> > 4 lpbh. > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > The goal is to use the best one at end. > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > > > > >> > Gilles Caulier > > > >> > > > > >> > > > > >> > > > > >> > 2018-01-04 13:18 GMT+01:00 Gilles Caulier > > > > <[hidden email] <mailto:[hidden email]>>: > > > >> > > The 6.0.0 git branch is created in remote : > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > [gilles@localhost digikam.master]$ git branch -a > > > >> > > > > > >> > > > > > >> > > > > > >> > > * master > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/HEAD -> origin/master > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/0.10 > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/0.8 > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/0.8.x > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/0.9.x > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/1.2 > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/1.3 > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/3.0.0 > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/6.0.0 < > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > ============================= > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/balooport > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/bug309058 > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/cmakecleanup > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/externaltools > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/frameworks > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/garbagecollection > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/hupnp > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/imagemetadatacache > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/kde3 > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/photolayoutseditor > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/sql2.0 > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/development/usbimport > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc11-clone > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc11-presentation > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc13-editorcanvas > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc13-imgqsort > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc13-localadj > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc13-tagmngr > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc13-thumbbar-mv > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc13-videow > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc14-greyc > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc14-kipibqm > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc14-qaccess > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc15-advmetadata > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc15-kio-drop > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc16-mysql > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc16-red-eye > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc17-db-sep > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc17-dlna > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc17-face-mngmnt > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/gsoc17-healing > > > >> > > > > > >> > > > > > >> > > > > > >> > > remotes/origin/master > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > This one is based on gsoc17-face-mngmnt sync with master > > > > today > > > > > >> morning. > > > >> > > > >> > > The remotes/origin/development/6.0.0 must be used to put > > > > next 6.0.0 > > > > > >> new > > > >> > > > >> > > code and must be sync with master until 5.x serie will be > > > > completed. > > > > > >> > > Best > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > > > > >> > > Gilles Caulier > > > >> > > > > > >> > > > > > >> > > > > > >> > > 2018-01-04 12:24 GMT+01:00 Gilles Caulier > > > > <[hidden email] <mailto:[hidden email]>>: > > > >> > >> The branch gsoc17-face-mngmnt is now sync with master > > > > and compile > > > > > >> > >> and > > > >> > >> > > > >> > >> > > > >> > >> > > > >> > >> link properly. > > > >> > >> > > > >> > >> > > > >> > >> > > > >> > >> > > > >> > >> > > > >> > >> > > > >> > >> > > > >> > >> I will create a new 6.0.0 branch based on the > > > > gsoc17-face-mngmnt > > > > > >> branch > > > >> > > > >> > >> for future main development. > > > >> > >> > > > >> > >> > > > >> > >> > > > >> > >> > > > >> > >> > > > >> > >> > > > >> > >> > > > >> > >> Gilles Caulier > > > >> > >> > > > >> > >> > > > >> > >> > > > >> > >> 2018-01-03 19:57 GMT+01:00 Maik Qualmann > > > > <[hidden email] <mailto:[hidden email]>>: > > > >> > >>> Gilles, and everyone else a Happy New Year. Yes, I'm > > > > looking > > > > > >> forward to > > > >> > > > >> > >>> feedback on MySQL - but only positive. ((:-)) No, > > > > seriously, I know > > > > > >> that > > > >> > > > >> > >>> there > > > >> > >>> > > > >> > >>> > > > >> > >>> > > > >> > >>> will be problems. Currently I only use MySQL for > > > > testing and it is > > > > > >> > >>> running > > > >> > >>> > > > >> > >>> > > > >> > >>> > > > >> > >>> really well here. It is important that we write in the > > > > release note > > > > > >> that > > > >> > > > >> > >>> the > > > >> > >>> > > > >> > >>> > > > >> > >>> > > > >> > >>> core DB with the migration tool should be copied into a > > > > new DB and > > > > > >> > >>> should be > > > >> > >>> > > > >> > >>> > > > >> > >>> > > > >> > >>> started with a new created thumbnail and face DB. > > > >> > >>> > > > >> > >>> > > > >> > >>> > > > >> > >>> > > > >> > >>> > > > >> > >>> > > > >> > >>> > > > >> > >>> Maik > > > >> > >>> > > > >> > >>> > > > >> > >>> > > > >> > >>> Am Mittwoch, 3. Januar 2018, 18:10:50 CET schrieb > > > > Gilles Caulier: > > > >> > >>> > Hi all, > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > In first, i hope an happy new year to all. > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > 1/ I currently revival the face management git branch > > > > written > > > > > >> while > > > >> > > > >> > >>> summer > > > >> > >>> > > > >> > >>> > 2017 by the student working on AI engine to recognize > > > > faces. The > > > > > >> code > > > >> > > > >> > >>> is > > > >> > >>> > > > >> > >>> > sync with master, but do not compile yet. I will fix it. > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > 2/ Important point : this code do not compile with > > > > OpenCV 2.x. It > > > > > >> > >>> require > > > >> > >>> > > > >> > >>> > OpenCV 3.1. This want mean that digiKam 6.0.0 will > > > > drop OpenCV > > > > > >> > >>> > 2.x > > > >> > >>> > > > >> > >>> support > > > >> > >>> > > > >> > >>> > definitively. > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > 3/ when faces management branch will be fixed, I > > > > would to create > > > > > >> a new > > > >> > > > >> > >>> one > > > >> > >>> > > > >> > >>> > dedicated to 6.0.0 release and let's master for the > > > > 5.x series, > > > > > >> > >>> > as > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > certainly a new 5.9.0 will be published few weeks ago > > > > 5.8.0, as > > > > > >> > >>> > maintenance, until 6.0.0 code will be stabilized. > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > 4/ The clone tool : this one still not very unusable > > > > for 5.8.0. I > > > > > >> > >>> recommend > > > >> > >>> > > > >> > >>> > to disable this option in editor for the moment, > > > > until zoom > > > > > >> support > > > >> > > > >> > >>> > and > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > brushes size preview over canvas is not implemented. > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > 5/ I do know yet when 5.8.0 will be released, as i > > > > don't receive > > > > > >> a new > > > >> > > > >> > >>> > response from Nicolas to create the whole tarball. > > > > Also Maik, > > > > > >> > >>> > want > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > certainly receive more feedback about Mysql support > > > > as i can see > > > > > >> > >>> recently > > > >> > >>> > > > >> > >>> > in bugzilla. > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > Best > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > > > > >> > >>> > Gilles Caulier > > > >> > > > >> -- > > > >> > > > >> > > > >> > > > >> Gruß Maik |
|
|
Ok, I found the problem: In tensor.h two serialize functions are
commented - no idea why. When uncommented they work fine. I'll commit in a second. On 11/01/18 19:14, Maik Qualmann wrote: > Yes, here also GCC 7.2.1. At the moment I am lost here too... > > Maik > > Am Donnerstag, 11. Januar 2018, 17:29:34 CET schrieb Simon Frei: >> I have the same problem, the first error I get is: >> >> In file included from >> /media/nunkystorage/4_Simon/Programme/Linux/source/digikam/develsoftwarecomp >> /core/libs/facesengine/facedb/facedb.cpp:28:0: >> /media/nunkystorage/4_Simon/Programme/Linux/source/digikam/develsoftwarecom >> p/core/libs/facesengine/dnnface/nn/layers.h: In function 'void >> serialize(const con_<_num_filters, _nr, _nc, >> _stride_y, _stride_x, _padding_y, _padding_x>&, std::ostream&)': >> /media/nunkystorage/4_Simon/Programme/Linux/source/digikam/develsoftwarecomp >> /core/libs/facesengine/dnnface/nn/layers.h:210:39: error: no matching >> function for call to 'serialize(const >> resizable_tensor&, std::ostream&)' >> serialize(item.params, out); >> ^ >> Then lots of candidates for serialize functions, but there is no >> serialize function defined for "const resizable_tensor&" as argument. So >> I am a bit confused why it compiles for you at all. Just in case, I am >> on gcc version 7.2. >> >> On 11/01/18 16:14, Gilles Caulier wrote: >>> Hi Maik, >>> >>> none. all compile fine here without extra package. >>> >>> Which GCC version do you use exactly ? >>> >>> Gilles >>> >>> 2018-01-11 12:41 GMT+01:00 Maik Qualmann <[hidden email] >>> >>> <mailto:[hidden email]>>: >>> Gilles, are any additional developer packages needed? I have many >>> notes here and then errors in: >>> >>> >>> >>> core/libs/facesengine/dnnface/dnn_base/serialize.h >>> >>> >>> >>> That's just a small excerpt: >>> >>> >>> >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dn >>> nface/nn/../matrix/../dnn_base/serialize.h:443:21:note: candidate: >>> void serialize(const short u >>> nsigned int&, std::ostream&) >>> inline void serialize (const T& item, std::ostream& out) \ >>> ^ >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dn >>> nface/nn/../matrix/../dnn_base/serialize.h:504:5:note: in expansion of >>> macro ‘USE_DEFAULT_INT_S >>> ERIALIZATION_FOR’ >>> USE_DEFAULT_INT_SERIALIZATION_FOR(unsigned short) >>> ^ >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dn >>> nface/nn/../matrix/../dnn_base/serialize.h:443:21:note: no known >>> conversion for argument 1 fr >>> om ‘const alias_tensor’ to ‘const short unsigned int&’ >>> inline void serialize (const T& item, std::ostream& out) \ >>> >>> In file included from >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/fa >>> cedb/facedb.cpp:28:0: >>> >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dn >>> nface/nn/layers.h:1162:39:error: no matching function for call to >>> ‘serialize(const alias_tensor >>> &, std::ostream&)’ >>> serialize(item.biases, out); >>> >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dn >>> nface/nn/../matrix/../dnn_base/serialize.h:1596:54:note: template >>> argument deduction/substitu >>> tion failed: >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dn >>> nface/nn/../matrix/../dnn_base/serialize.h: In substitution of >>> ‘template<class T> typename enabl >>> e_if<is_protocol_buffer<T> >::type serialize(const T&, >>> std::ostream&) [with T = alias_tensor]’: >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dn >>> nface/nn/layers.h:1162:39: required from here >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/dn >>> nface/nn/../matrix/../dnn_base/serialize.h:1596:54:error: no type >>> named ‘type’ in ‘struct enabl >>> e_if<is_protocol_buffer<alias_tensor, void>, void>’ >>> >>> >>> >>> Maik >>> >>> Am Sonntag, 7. Januar 2018, 20:05:50 CET schrieb Gilles Caulier: >>> > There is nothing more in error trace, as a pure method is not >>> >>> implemented >>> >>> > in EigenFaceRecognizer ? >>> > >>> > >>> > >>> > Gilles >>> > >>> > >>> > >>> > 2018-01-07 20:02 GMT+01:00 Gilles Caulier >>> >>> <[hidden email] <mailto:[hidden email]>>: >>> > > Hum all compile fine here with OPenCV 3.4. Which version do >>> >>> you use ? >>> >>> > > Gilles >>> > > >>> > > >>> > > >>> > > 2018-01-07 19:55 GMT+01:00 Maik Qualmann >>> >>> <[hidden email] <mailto:[hidden email]>>: >>> > >> Compile error: >>> > >> >>> > >> >>> > >> >>> > >> >>> > >> >>> > >> >>> > >> >>> > >> /home/maik/Devel/digikam-software-compilation/core/libs/ >>> >>> facesengine/recognition-opencv-eigenfaces/facerec_eigenborrowed.cpp: >>> In >>> >>> > >> static member function ‘static cv::P >>> > >> >>> > >> tr<Digikam::EigenFaceRecognizer> Digikam::EigenFaceRecognizer:: >>> > >> >>> > >> create(double)’: >>> > >> >>> > >> /home/maik/Devel/digikam-software-compilation/core/libs/ >>> > >> >>> > >> facesengine/recognition-opencv-eigenfaces/facerec_eigenborro >>> > >> >>> > >> wed.cpp:216:70: error: invalid new-expression of >>> > >> >>> > >> abstract class type ‘Digikam::EigenFaceRecognizer’ >>> > >> >>> > >> >>> > >> >>> > >> EigenFaceRecognizer* const fr = new >>> >>> EigenFaceRecognizer(threshold); >>> >>> > >> ^ >>> > >> >>> > >> >>> > >> >>> > >> Maik >>> > >> >>> > >> Am Samstag, 6. Januar 2018, 17:53:30 CET schrieb Gilles Caulier: >>> > >> > The main integration and improvements of new face recognition >>> > >> >>> > >> algorithms in >>> > >> >>> > >> > 6.0.0 branch is now complete and ready to use. >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > The default algorithm is DNN (deep neural network => deep >>> >>> learning => >>> >>> > >> IA). >>> > >> >>> > >> > In face scan dialog, in advanced tab, we have a new option >>> >>> where >>> >>> > >> > recognition algorithm can be selected. >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > - lpbh : older one present on 5.x based on OpenCV ( >>> >>> https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face >>> <https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face> >>> >>> > >> rec_tutorial.htm >>> > >> >>> > >> > l#local-binary-patterns-histograms ) >>> > >> > >>> > >> > >>> > >> > >>> > >> > - fisherfaces : OpenCV based ( >>> >>> https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face >>> <https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face> >>> >>> > >> rec_tutorial.htm >>> > >> >>> > >> > l#fisherfaces ) >>> > >> > >>> > >> > >>> > >> > >>> > >> > - eigenfaces : OpenCV based ( >>> >>> https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face >>> <https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face> >>> >>> > >> rec_tutorial.htm >>> > >> >>> > >> > l#eigenfaces ) >>> > >> > >>> > >> > >>> > >> > >>> > >> > - dnn : dlib based ( >>> >>> http://blog.dlib.net/2017/02/high-quality-face-recognition-w >>> <http://blog.dlib.net/2017/02/high-quality-face-recognition-w> >>> >>> > >> ith-deep.html) >>> > >> >>> > >> > I'm not yet satisfied by DNN. I'm not started to compare >>> >>> the algorithm >>> >>> > >> with >>> > >> >>> > >> > a clean database. In all cases, we must detect all faces, >>> >>> not tagged. >>> >>> > >> After >>> > >> >>> > >> > that, we need to train the faces with names, else the >>> >>> recognition will >>> >>> > >> not >>> > >> >>> > >> > work. As i know, at least with lpbh algorithm, we need at >>> >>> least 6 >>> >>> > >> > images >>> > >> > >>> > >> > >>> > >> > >>> > >> > tagged with same name to see a recognition result while >>> >>> scanning. To >>> >>> > >> have a >>> > >> >>> > >> > clean compare result, for each algorithm to use, we need a >>> >>> clear >>> >>> > >> database. >>> > >> >>> > >> > The student working on this project while summer 2017 >>> >>> report to me that >>> >>> > >> in >>> > >> >>> > >> > all cases the best algorithms are : >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > 1 DNN. >>> > >> > >>> > >> > >>> > >> > >>> > >> > 2 eigenfaces. >>> > >> > >>> > >> > >>> > >> > >>> > >> > 3 fisherfaces. >>> > >> > >>> > >> > >>> > >> > >>> > >> > 4 lpbh. >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > The goal is to use the best one at end. >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > Gilles Caulier >>> > >> > >>> > >> > >>> > >> > >>> > >> > 2018-01-04 13:18 GMT+01:00 Gilles Caulier >>> >>> <[hidden email] <mailto:[hidden email]>>: >>> > >> > > The 6.0.0 git branch is created in remote : >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > [gilles@localhost digikam.master]$ git branch -a >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > * master >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/HEAD -> origin/master >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/0.10 >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/0.8 >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/0.8.x >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/0.9.x >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/1.2 >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/1.3 >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/3.0.0 >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/6.0.0 < >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > ============================= >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/balooport >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/bug309058 >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/cmakecleanup >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/externaltools >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/frameworks >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/garbagecollection >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/hupnp >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/imagemetadatacache >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/kde3 >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/photolayoutseditor >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/sql2.0 >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/development/usbimport >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc11-clone >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc11-presentation >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc13-editorcanvas >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc13-imgqsort >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc13-localadj >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc13-tagmngr >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc13-thumbbar-mv >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc13-videow >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc14-greyc >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc14-kipibqm >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc14-qaccess >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc15-advmetadata >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc15-kio-drop >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc16-mysql >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc16-red-eye >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc17-db-sep >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc17-dlna >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc17-face-mngmnt >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/gsoc17-healing >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > remotes/origin/master >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > This one is based on gsoc17-face-mngmnt sync with master >>> >>> today >>> >>> > >> morning. >>> > >> >>> > >> > > The remotes/origin/development/6.0.0 must be used to put >>> >>> next 6.0.0 >>> >>> > >> new >>> > >> >>> > >> > > code and must be sync with master until 5.x serie will be >>> >>> completed. >>> >>> > >> > > Best >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > Gilles Caulier >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > 2018-01-04 12:24 GMT+01:00 Gilles Caulier >>> >>> <[hidden email] <mailto:[hidden email]>>: >>> > >> > >> The branch gsoc17-face-mngmnt is now sync with master >>> >>> and compile >>> >>> > >> > >> and >>> > >> > >> >>> > >> > >> >>> > >> > >> >>> > >> > >> link properly. >>> > >> > >> >>> > >> > >> >>> > >> > >> >>> > >> > >> >>> > >> > >> >>> > >> > >> >>> > >> > >> >>> > >> > >> I will create a new 6.0.0 branch based on the >>> >>> gsoc17-face-mngmnt >>> >>> > >> branch >>> > >> >>> > >> > >> for future main development. >>> > >> > >> >>> > >> > >> >>> > >> > >> >>> > >> > >> >>> > >> > >> >>> > >> > >> >>> > >> > >> >>> > >> > >> Gilles Caulier >>> > >> > >> >>> > >> > >> >>> > >> > >> >>> > >> > >> 2018-01-03 19:57 GMT+01:00 Maik Qualmann >>> >>> <[hidden email] <mailto:[hidden email]>>: >>> > >> > >>> Gilles, and everyone else a Happy New Year. Yes, I'm >>> >>> looking >>> >>> > >> forward to >>> > >> >>> > >> > >>> feedback on MySQL - but only positive. ((:-)) No, >>> >>> seriously, I know >>> >>> > >> that >>> > >> >>> > >> > >>> there >>> > >> > >>> >>> > >> > >>> >>> > >> > >>> >>> > >> > >>> will be problems. Currently I only use MySQL for >>> >>> testing and it is >>> >>> > >> > >>> running >>> > >> > >>> >>> > >> > >>> >>> > >> > >>> >>> > >> > >>> really well here. It is important that we write in the >>> >>> release note >>> >>> > >> that >>> > >> >>> > >> > >>> the >>> > >> > >>> >>> > >> > >>> >>> > >> > >>> >>> > >> > >>> core DB with the migration tool should be copied into a >>> >>> new DB and >>> >>> > >> > >>> should be >>> > >> > >>> >>> > >> > >>> >>> > >> > >>> >>> > >> > >>> started with a new created thumbnail and face DB. >>> > >> > >>> >>> > >> > >>> >>> > >> > >>> >>> > >> > >>> >>> > >> > >>> >>> > >> > >>> >>> > >> > >>> >>> > >> > >>> Maik >>> > >> > >>> >>> > >> > >>> >>> > >> > >>> >>> > >> > >>> Am Mittwoch, 3. Januar 2018, 18:10:50 CET schrieb >>> >>> Gilles Caulier: >>> > >> > >>> > Hi all, >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > In first, i hope an happy new year to all. >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > 1/ I currently revival the face management git branch >>> >>> written >>> >>> > >> while >>> > >> >>> > >> > >>> summer >>> > >> > >>> >>> > >> > >>> > 2017 by the student working on AI engine to recognize >>> >>> faces. The >>> >>> > >> code >>> > >> >>> > >> > >>> is >>> > >> > >>> >>> > >> > >>> > sync with master, but do not compile yet. I will fix it. >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > 2/ Important point : this code do not compile with >>> >>> OpenCV 2.x. It >>> >>> > >> > >>> require >>> > >> > >>> >>> > >> > >>> > OpenCV 3.1. This want mean that digiKam 6.0.0 will >>> >>> drop OpenCV >>> >>> > >> > >>> > 2.x >>> > >> > >>> >>> > >> > >>> support >>> > >> > >>> >>> > >> > >>> > definitively. >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > 3/ when faces management branch will be fixed, I >>> >>> would to create >>> >>> > >> a new >>> > >> >>> > >> > >>> one >>> > >> > >>> >>> > >> > >>> > dedicated to 6.0.0 release and let's master for the >>> >>> 5.x series, >>> >>> > >> > >>> > as >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > certainly a new 5.9.0 will be published few weeks ago >>> >>> 5.8.0, as >>> >>> > >> > >>> > maintenance, until 6.0.0 code will be stabilized. >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > 4/ The clone tool : this one still not very unusable >>> >>> for 5.8.0. I >>> >>> > >> > >>> recommend >>> > >> > >>> >>> > >> > >>> > to disable this option in editor for the moment, >>> >>> until zoom >>> >>> > >> support >>> > >> >>> > >> > >>> > and >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > brushes size preview over canvas is not implemented. >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > 5/ I do know yet when 5.8.0 will be released, as i >>> >>> don't receive >>> >>> > >> a new >>> > >> >>> > >> > >>> > response from Nicolas to create the whole tarball. >>> >>> Also Maik, >>> >>> > >> > >>> > want >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > certainly receive more feedback about Mysql support >>> >>> as i can see >>> >>> > >> > >>> recently >>> > >> > >>> >>> > >> > >>> > in bugzilla. >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > Best >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > Gilles Caulier >>> > >> >>> > >> -- >>> > >> >>> > >> >>> > >> >>> > >> Gruß Maik > |
|
|
Yes, thanks, that's it...
Maik Am Donnerstag, 11. Januar 2018, 19:44:37 CET schrieb Simon Frei: > Ok, I found the problem: In tensor.h two serialize functions are > commented - no idea why. When uncommented they work fine. I'll commit in > a second. > > On 11/01/18 19:14, Maik Qualmann wrote: > > Yes, here also GCC 7.2.1. At the moment I am lost here too... > > > > Maik > > > > Am Donnerstag, 11. Januar 2018, 17:29:34 CET schrieb Simon Frei: > >> I have the same problem, the first error I get is: > >> > >> In file included from > >> /media/nunkystorage/4_Simon/Programme/Linux/source/digikam/develsoftwarec > >> omp /core/libs/facesengine/facedb/facedb.cpp:28:0: > >> /media/nunkystorage/4_Simon/Programme/Linux/source/digikam/develsoftwarec > >> om > >> p/core/libs/facesengine/dnnface/nn/layers.h: In function 'void > >> serialize(const con_<_num_filters, _nr, _nc, > >> _stride_y, _stride_x, _padding_y, _padding_x>&, std::ostream&)': > >> /media/nunkystorage/4_Simon/Programme/Linux/source/digikam/develsoftwarec > >> omp /core/libs/facesengine/dnnface/nn/layers.h:210:39: error: no matching > >> function for call to 'serialize(const > >> resizable_tensor&, std::ostream&)' > >> > >> serialize(item.params, out); > >> > >> ^ > >> > >> Then lots of candidates for serialize functions, but there is no > >> serialize function defined for "const resizable_tensor&" as argument. So > >> I am a bit confused why it compiles for you at all. Just in case, I am > >> on gcc version 7.2. > >> > >> On 11/01/18 16:14, Gilles Caulier wrote: > >>> Hi Maik, > >>> > >>> none. all compile fine here without extra package. > >>> > >>> Which GCC version do you use exactly ? > >>> > >>> Gilles > >>> > >>> 2018-01-11 12:41 GMT+01:00 Maik Qualmann <[hidden email] > >>> > >>> <mailto:[hidden email]>>: > >>> Gilles, are any additional developer packages needed? I have many > >>> notes here and then errors in: > >>> > >>> > >>> > >>> core/libs/facesengine/dnnface/dnn_base/serialize.h > >>> > >>> > >>> > >>> That's just a small excerpt: > >>> > >>> > >>> > >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/ > >>> dn > >>> nface/nn/../matrix/../dnn_base/serialize.h:443:21:note: candidate: > >>> void serialize(const short u > >>> nsigned int&, std::ostream&) > >>> > >>> inline void serialize (const T& item, std::ostream& out) \ > >>> > >>> ^ > >>> > >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/ > >>> dn > >>> nface/nn/../matrix/../dnn_base/serialize.h:504:5:note: in expansion > >>> of > >>> macro ‘USE_DEFAULT_INT_S > >>> ERIALIZATION_FOR’ > >>> > >>> USE_DEFAULT_INT_SERIALIZATION_FOR(unsigned short) > >>> ^ > >>> > >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/ > >>> dn > >>> nface/nn/../matrix/../dnn_base/serialize.h:443:21:note: no known > >>> conversion for argument 1 fr > >>> om ‘const alias_tensor’ to ‘const short unsigned int&’ > >>> > >>> inline void serialize (const T& item, std::ostream& out) \ > >>> > >>> In file included from > >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/ > >>> fa > >>> cedb/facedb.cpp:28:0: > >>> > >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/ > >>> dn > >>> nface/nn/layers.h:1162:39:error: no matching function for call to > >>> ‘serialize(const alias_tensor > >>> &, std::ostream&)’ > >>> > >>> serialize(item.biases, out); > >>> > >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/ > >>> dn > >>> nface/nn/../matrix/../dnn_base/serialize.h:1596:54:note: template > >>> argument deduction/substitu > >>> tion failed: > >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/ > >>> dn > >>> nface/nn/../matrix/../dnn_base/serialize.h: In substitution of > >>> ‘template<class T> typename enabl > >>> e_if<is_protocol_buffer<T> >::type serialize(const T&, > >>> std::ostream&) [with T = alias_tensor]’: > >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/ > >>> dn > >>> nface/nn/layers.h:1162:39: required from here > >>> /home/maik/Devel/digikam-software-compilation/core/libs/facesengine/ > >>> dn > >>> nface/nn/../matrix/../dnn_base/serialize.h:1596:54:error: no type > >>> named ‘type’ in ‘struct enabl > >>> e_if<is_protocol_buffer<alias_tensor, void>, void>’ > >>> > >>> > >>> > >>> Maik > >>> > >>> Am Sonntag, 7. Januar 2018, 20:05:50 CET schrieb Gilles Caulier: > >>> > There is nothing more in error trace, as a pure method is not > >>> > >>> implemented > >>> > >>> > in EigenFaceRecognizer ? > >>> > > >>> > > >>> > > >>> > Gilles > >>> > > >>> > > >>> > > >>> > 2018-01-07 20:02 GMT+01:00 Gilles Caulier > >>> > >>> <[hidden email] <mailto:[hidden email]>>: > >>> > > Hum all compile fine here with OPenCV 3.4. Which version do > >>> > >>> you use ? > >>> > >>> > > Gilles > >>> > > > >>> > > > >>> > > > >>> > > 2018-01-07 19:55 GMT+01:00 Maik Qualmann > >>> > >>> <[hidden email] <mailto:[hidden email]>>: > >>> > >> Compile error: > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> > >>> > >> /home/maik/Devel/digikam-software-compilation/core/libs/ > >>> > >>> facesengine/recognition-opencv-eigenfaces/facerec_eigenborrowed.cpp: > >>> In > >>> > >>> > >> static member function ‘static cv::P > >>> > >> > >>> > >> tr<Digikam::EigenFaceRecognizer> Digikam::EigenFaceRecognizer:: > >>> > >> > >>> > >> create(double)’: > >>> > >> > >>> > >> /home/maik/Devel/digikam-software-compilation/core/libs/ > >>> > >> > >>> > >> facesengine/recognition-opencv-eigenfaces/facerec_eigenborro > >>> > >> > >>> > >> wed.cpp:216:70: error: invalid new-expression of > >>> > >> > >>> > >> abstract class type ‘Digikam::EigenFaceRecognizer’ > >>> > >> > >>> > >> > >>> > >> > >>> > >> EigenFaceRecognizer* const fr = new > >>> > >>> EigenFaceRecognizer(threshold); > >>> > >>> > >> ^ > >>> > >> > >>> > >> > >>> > >> > >>> > >> Maik > >>> > >> > >>> > >> Am Samstag, 6. Januar 2018, 17:53:30 CET schrieb Gilles > >>> > >> > The main integration and improvements of new face recognition > >>> > >> > >>> > >> algorithms in > >>> > >> > >>> > >> > 6.0.0 branch is now complete and ready to use. > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > The default algorithm is DNN (deep neural network => deep > >>> > >>> learning => > >>> > >>> > >> IA). > >>> > >> > >>> > >> > In face scan dialog, in advanced tab, we have a new option > >>> > >>> where > >>> > >>> > >> > recognition algorithm can be selected. > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > - lpbh : older one present on 5.x based on OpenCV ( > >>> > >>> https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face > >>> <https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face> > >>> > >>> > >> rec_tutorial.htm > >>> > >> > >>> > >> > l#local-binary-patterns-histograms ) > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > - fisherfaces : OpenCV based ( > >>> > >>> https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face > >>> <https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face> > >>> > >>> > >> rec_tutorial.htm > >>> > >> > >>> > >> > l#fisherfaces ) > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > - eigenfaces : OpenCV based ( > >>> > >>> https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face > >>> <https://docs.opencv.org/2.4/modules/contrib/doc/facerec/face> > >>> > >>> > >> rec_tutorial.htm > >>> > >> > >>> > >> > l#eigenfaces ) > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > - dnn : dlib based ( > >>> > >>> http://blog.dlib.net/2017/02/high-quality-face-recognition-w > >>> <http://blog.dlib.net/2017/02/high-quality-face-recognition-w> > >>> > >>> > >> ith-deep.html) > >>> > >> > >>> > >> > I'm not yet satisfied by DNN. I'm not started to compare > >>> > >>> the algorithm > >>> > >>> > >> with > >>> > >> > >>> > >> > a clean database. In all cases, we must detect all faces, > >>> > >>> not tagged. > >>> > >>> > >> After > >>> > >> > >>> > >> > that, we need to train the faces with names, else the > >>> > >>> recognition will > >>> > >>> > >> not > >>> > >> > >>> > >> > work. As i know, at least with lpbh algorithm, we need at > >>> > >>> least 6 > >>> > >>> > >> > images > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > tagged with same name to see a recognition result while > >>> > >>> scanning. To > >>> > >>> > >> have a > >>> > >> > >>> > >> > clean compare result, for each algorithm to use, we need a > >>> > >>> clear > >>> > >>> > >> database. > >>> > >> > >>> > >> > The student working on this project while summer 2017 > >>> > >>> report to me that > >>> > >>> > >> in > >>> > >> > >>> > >> > all cases the best algorithms are : > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > 1 DNN. > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > 2 eigenfaces. > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > 3 fisherfaces. > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > 4 lpbh. > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > The goal is to use the best one at end. > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > Gilles Caulier > >>> > >> > > >>> > >> > > >>> > >> > > >>> > >> > 2018-01-04 13:18 GMT+01:00 Gilles Caulier > >>> > >>> <[hidden email] <mailto:[hidden email]>>: > >>> > >> > > The 6.0.0 git branch is created in remote : > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > [gilles@localhost digikam.master]$ git branch -a > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > * master > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/HEAD -> origin/master > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/0.10 > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/0.8 > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/0.8.x > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/0.9.x > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/1.2 > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/1.3 > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/3.0.0 > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/6.0.0 < > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > ============================= > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/balooport > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/bug309058 > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/cmakecleanup > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/externaltools > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/frameworks > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/garbagecollection > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/hupnp > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/imagemetadatacache > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/kde3 > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/photolayoutseditor > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/sql2.0 > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/development/usbimport > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc11-clone > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc11-presentation > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc13-editorcanvas > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc13-imgqsort > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc13-localadj > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc13-tagmngr > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc13-thumbbar-mv > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc13-videow > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc14-greyc > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc14-kipibqm > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc14-qaccess > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc15-advmetadata > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc15-kio-drop > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc16-mysql > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc16-red-eye > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc17-db-sep > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc17-dlna > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc17-face-mngmnt > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/gsoc17-healing > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > remotes/origin/master > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > This one is based on gsoc17-face-mngmnt sync with master > >>> > >>> today > >>> > >>> > >> morning. > >>> > >> > >>> > >> > > The remotes/origin/development/6.0.0 must be used to put > >>> > >>> next 6.0.0 > >>> > >>> > >> new > >>> > >> > >>> > >> > > code and must be sync with master until 5.x serie will be > >>> > >>> completed. > >>> > >>> > >> > > Best > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > Gilles Caulier > >>> > >> > > > >>> > >> > > > >>> > >> > > > >>> > >> > > 2018-01-04 12:24 GMT+01:00 Gilles Caulier > >>> > >>> <[hidden email] <mailto:[hidden email]>>: > >>> > >> > >> The branch gsoc17-face-mngmnt is now sync with master > >>> > >>> and compile > >>> > >>> > >> > >> and > >>> > >> > >> > >>> > >> > >> > >>> > >> > >> > >>> > >> > >> link properly. > >>> > >> > >> > >>> > >> > >> > >>> > >> > >> > >>> > >> > >> > >>> > >> > >> > >>> > >> > >> > >>> > >> > >> > >>> > >> > >> I will create a new 6.0.0 branch based on the > >>> > >>> gsoc17-face-mngmnt > >>> > >>> > >> branch > >>> > >> > >>> > >> > >> for future main development. > >>> > >> > >> > >>> > >> > >> > >>> > >> > >> > >>> > >> > >> > >>> > >> > >> > >>> > >> > >> > >>> > >> > >> > >>> > >> > >> Gilles Caulier > >>> > >> > >> > >>> > >> > >> > >>> > >> > >> > >>> > >> > >> 2018-01-03 19:57 GMT+01:00 Maik Qualmann > >>> > >>> <[hidden email] <mailto:[hidden email]>>: > >>> > >> > >>> Gilles, and everyone else a Happy New Year. Yes, I'm > >>> > >>> looking > >>> > >>> > >> forward to > >>> > >> > >>> > >> > >>> feedback on MySQL - but only positive. ((:-)) No, > >>> > >>> seriously, I know > >>> > >>> > >> that > >>> > >> > >>> > >> > >>> there > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> will be problems. Currently I only use MySQL for > >>> > >>> testing and it is > >>> > >>> > >> > >>> running > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> really well here. It is important that we write in the > >>> > >>> release note > >>> > >>> > >> that > >>> > >> > >>> > >> > >>> the > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> core DB with the migration tool should be copied into a > >>> > >>> new DB and > >>> > >>> > >> > >>> should be > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> started with a new created thumbnail and face DB. > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> Maik > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> > >>> > >> > >>> Am Mittwoch, 3. Januar 2018, 18:10:50 CET schrieb > >>> > >>> Gilles Caulier: > >>> > >> > >>> > Hi all, > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > In first, i hope an happy new year to all. > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > 1/ I currently revival the face management git branch > >>> > >>> written > >>> > >>> > >> while > >>> > >> > >>> > >> > >>> summer > >>> > >> > >>> > >>> > >> > >>> > 2017 by the student working on AI engine to recognize > >>> > >>> faces. The > >>> > >>> > >> code > >>> > >> > >>> > >> > >>> is > >>> > >> > >>> > >>> > >> > >>> > sync with master, but do not compile yet. I will fix > >>> > >> > >>> > it. > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > 2/ Important point : this code do not compile with > >>> > >>> OpenCV 2.x. It > >>> > >>> > >> > >>> require > >>> > >> > >>> > >>> > >> > >>> > OpenCV 3.1. This want mean that digiKam 6.0.0 will > >>> > >>> drop OpenCV > >>> > >>> > >> > >>> > 2.x > >>> > >> > >>> > >>> > >> > >>> support > >>> > >> > >>> > >>> > >> > >>> > definitively. > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > 3/ when faces management branch will be fixed, I > >>> > >>> would to create > >>> > >>> > >> a new > >>> > >> > >>> > >> > >>> one > >>> > >> > >>> > >>> > >> > >>> > dedicated to 6.0.0 release and let's master for the > >>> > >>> 5.x series, > >>> > >>> > >> > >>> > as > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > certainly a new 5.9.0 will be published few weeks ago > >>> > >>> 5.8.0, as > >>> > >>> > >> > >>> > maintenance, until 6.0.0 code will be stabilized. > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > 4/ The clone tool : this one still not very unusable > >>> > >>> for 5.8.0. I > >>> > >>> > >> > >>> recommend > >>> > >> > >>> > >>> > >> > >>> > to disable this option in editor for the moment, > >>> > >>> until zoom > >>> > >>> > >> support > >>> > >> > >>> > >> > >>> > and > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > brushes size preview over canvas is not implemented. > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > 5/ I do know yet when 5.8.0 will be released, as i > >>> > >>> don't receive > >>> > >>> > >> a new > >>> > >> > >>> > >> > >>> > response from Nicolas to create the whole tarball. > >>> > >>> Also Maik, > >>> > >>> > >> > >>> > want > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > certainly receive more feedback about Mysql support > >>> > >>> as i can see > >>> > >>> > >> > >>> recently > >>> > >> > >>> > >>> > >> > >>> > in bugzilla. > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > Best > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > > >>> > >> > >>> > Gilles Caulier > >>> > >> > >>> > >> -- > >>> > >> > >>> > >> > >>> > >> > >>> > >> Gruß Maik |
| Free forum by Nabble | Edit this page |