search

cvg / pixloc

Back to the Feature: Learning Robust Camera Localization from Pixels to Pose (CVPR 2021)
Apache License 2.0
735 stars 92 forks source link

Using COLMAPS estimated and refined camera intrinsics as input for localization? #51

Open tworkool opened 2 months ago

tworkool commented 2 months ago

Hello, As the documentation states, Pixloc requires a text file with cameras and their intrinsics. I am working in a project which has one camera with unknown intrinsics among others which have intrinsics. Is it possible to use COLMAP's estimated camera intrinsics in that camera file? The focal length in particular is what I worry about because it is not up to scale...

paths = paths.add_prefixes(DATA_PATH , LOC_PATH)
paths.query_list = DATA_PATH / dataset / "queries_with_intrinsics.txt" #<----

localizer = PoseLocalizer(paths, conf)