Open QuangVinh1280 opened 5 months ago
You can train LightGlue on just a set of images with homography adaptation. The best training requires posed images with depth maps. This can, for example, be generated with COLMAP SfM + MVS.
You can train LightGlue on just a set of images with homography adaptation. The best training requires posed images with depth maps. This can, for example, be generated with COLMAP SfM + MVS.
Thank you for your feedback. So, to retrain LightGlue, I need images with pose and depth maps, right? Could you provide more information about the required format of the data
Hello everyone, I am currently studying to retrain the LightGlue model, but right now I don't know what information I need to proceed with retraining the LightGlue model. Could anyone explain to me? Thank you very much