Does this toolbox support localization of a new image into an existing SfM model constructed using default COLMAP?

[molegod]

I'm new to this area and using COLMAP generally, so I'm a bit unsure! I looked at the demos, but from what I can tell localization needs a text file listing out covisible images? Do I need to run something like NetVLAD every time I add a new image to generate a new text file?

[swengeler]

If default COLMAP is used for the reference reconstruction (using SIFT features and none of the feature extraction/matching in hloc), would registering/localizing the new image using default COLMAP be sufficient or is there something else you're trying to use hloc for specifically? There's a section in the COLMAP FAQ to do exactly that, which seems like it would be the easiest way to achieve what you're asking about.

[JiajieLi7012]

If default COLMAP is used for the reference reconstruction (using SIFT features and none of the feature extraction/matching in hloc), would registering/localizing the new image using default COLMAP be sufficient or is there something else you're trying to use hloc for specifically? There's a section in the COLMAP FAQ to do exactly that, which seems like it would be the easiest way to achieve what you're asking about.

Hi, I'm quite new to this field and I'm also wondering if it's possible to localize a set of new images using superpoint + superglue with hloc against a SfM model constructed by default COLMAP. Any help would be appreciated. Thanks in advance.