Project Name:
digiKam: Face Management Improvements
Brief Description:
Face recognition in digiKam is now implemented using Local Binary Patterns Histograms (LBPH) from OpenCV. However, since the performance is not as expected, the algorithm needs to be improved. To improve face recognition, new face algorithms with pose estimation and normalization will be added. The algorithms will be selectable in GUI by users. Besides the algorithms, the face region in the database should be synchronized when the image is transformed.
Finished Work:
1. Eigenfaces and Fisherfaces algorithm has been added to digiKam. The accuracy of face recognition has been improved to 80%.
2. Deep learning algorithm has been added into digiKam. The accuracy of face recognition has been improved to 99%.
3. The face algorithms is selectable in GUI by users.
4. The face tag region in preview mode is synchronized between database and metadata when rotating an image.
5. The face tag region in Image Editor is synchronized between database and metadata when rotating or flipping an image.
Future Work:
1. Add undo and redo in Image Editor for image transformation.
2. Add face tag region synchronization when image is cropped or resized.
Project Commits:
https://cgit.kde.org/digikam.git/log/?h=gsoc17-face-mngmnt&qt=author&q=Yingjie
Status Report:
https://community.kde.org/GSoC/2017/StatusReports/YingjieLiu
List of Fixed Bugs:
https://bugs.kde.org/show_bug.cgi?id=376681
https://bugs.kde.org/show_bug.cgi?id=326538
https://bugs.kde.org/show_bug.cgi?id=381378