[WIP] Describer types brisk, surf (new, sorted)

[natowi]

Description

Added brisk_ocv and surf_ocv describers for https://github.com/alicevision/AliceVision/pull/702 (waiting for merge) to FeatureExtraction, FeatureMatching, StructureFromMotion nodes.

ToDo: ExportMatches, ConvertSfMFormat, SfMTransform, GlobalSfM, PanoramaEstimation, CameraRigCalibration, CameraRigLocalization, CameraLocalization

Implementation remarks

I grouped the describers by native, ocv and cctag, sorted in alphabetical order:

values=['akaze', 'akaze_liop', 'akaze_mldb', 'sift', 'sift_float', 'sift_upright', 'akaze_ocv', 'brisk_ocv', 'sift_ocv', 'surf_ocv', 'cctag3', 'cctag4'],

Another logical order could be the following:

values=['sift', 'sift_ocv', 'sift_float', 'sift_upright', 'akaze', 'akaze_ocv', 'akaze_liop', 'akaze_mldb', 'brisk_ocv', 'surf_ocv', 'cctag3', 'cctag4'],

[simogasp]

A good idea would be to have some sort of general setting/variable that contains the list in order to avoid code replication in each concerned node (ie declare the list once and use it in each node). That is more maintainable than having the the list replicated in several parts of the code.

Also, I don't know if it is feasible but the list should only contain the supported features for the given alicevision binaries: I don't know if we can use the env variable to go see the config.hpp and check e.g. is opencv is supported and drop all relevant features if not. The same for cctag etc...

[fabiencastan]

Yes, good idea, we can declare a list in FeatureExtraction.py and import that from FeatureMatching and StructureFromMotion. To keep it simple, we will assume that opencv is available and this will be the case in the releases.

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[vic4key]

Please merge this feature @fabiencastan. P.S Sorry for pulling this thread up after a long time. But these algorithms might better for several cases. It should be merged.

[fabiencastan]

The only pbl afaik is that it needs to be finalized.