The datasets utilized in our experiments can be found on their websites, including scan data and their semantic annotations.
Keypoints and features can be extracted using D3Feat or other extraction methods.
Semantic segmentation results can be obtained using SphereFormer or other segmentation methods.
Code will be released after the paper is accepted.