Rescale colmap keypoints. #817
Conversation
|
Do the results when loading in the COLMAP database now align with what you get when running GTSfM with the COLMAP frontend (i.e., running with the |
@travisdriver Hi Travis, apologies for the delayed response; I haven’t checked this account in a while. Are you asking whether the results from running GTSfM with the COLMAP frontend align with the results from running the full end-to-end COLMAP pipeline? |
| ) | ||
| return Keypoints(coordinates=coordinates, scales=None, responses=None) | ||
|
|
||
| def _read_image_ids_and_keypoints( |
There was a problem hiding this comment.
are you using the same formatting settings? please refer to https://github.com/borglab/gtsfm/blob/master/CONTRIBUTING.md#python-style
| two_view_geometry = self._pycolmap_db.read_two_view_geometry(colmap_i1, colmap_i2) | ||
| two_view_geometry = self._pycolmap_db.read_two_view_geometry( | ||
| colmap_i1, colmap_i2 | ||
| ) |
There was a problem hiding this comment.
undesirable format change.
|
|
||
| # Only read matches if we have an essential or a fundamental matrix | ||
| if two_view_geometry.config != 2 and two_view_geometry.config != 3: | ||
| continue | ||
|
|
||
| # Note(Ayush): the matches we are loading are actually post verification | ||
| corr_idxs[(i1, i2)] = np.array(two_view_geometry.inlier_matches, dtype=np.int32) | ||
| corr_idxs[(i1, i2)] = np.array( |
There was a problem hiding this comment.
undesirable format change.
| @@ -107,7 +160,9 @@ def generate_correspondences( | |||
| # Note: we will end up reading verified correspondences from the colmap DB. | |||
| images_actual = client.gather(images) | |||
|
|
|||
| gtsfm_id_to_pycolmap_id, keypoints = self._read_image_ids_and_keypoints(images_actual) | |||
| gtsfm_id_to_pycolmap_id, keypoints = self._read_image_ids_and_keypoints( | |||
There was a problem hiding this comment.
undesirable format change.
The purpose of this PR is to enable running GTSFM with feature extraction results from COLMAP's front end. COLMAP front-end results are typically stored in a database format. The
colmap_correspondence_generator.pylibrary aim to load COLMAP's feature extraction results directly from this database.However, there is a discrepancy in how COLMAP handles feature extraction and storage:
Key Issues Addressed by this PR:
Important Notes: