-
Improvement: support image database update
-
Detailed unit test for each module
-
Deep learning methods support (tensorflow)
-
Hashing methods support
-
Re-organize code structure
- better cmake organization (add installation rules)
-
Add more feature support, such as VLFeat, BRIEF, etc.
- add vlfeat (SIFT)
-
benchmark dataset performance (2D / 3D)
-
SLAM-friendly development, add loop closure detection
-
Better engineering
- command line system (e.g. gflags)
- doxygen documentation
- python export
-
Add option to support opencv
- image I/O currently rely on opencv
- opencv feature support (contains non-free modules)