Happing today! 📣
I'll be presenting our work "CutS3D: Cutting Semantics in 3D for 2D Unsupervised Instance Segmentation" in the morning poster session at #ICCV2025, located at poster no. 136! 🌺 Come by and say aloha! 👋🏻 If you can't make it, check out our project page at leonsick.github.io/cuts3d/

Happy to share our new #SIGGRAPHAsia2025 paper! If you are tired of using marching cubes to extract your mesh from neural implicit representations, check out Marching Neurons!

Project: phermosilla.github.io/neurons/
Arxiv: arxiv.org/abs/2509.21007

Happy to share our recent work on hierarchical 3D scene understanding using hyperbolic spaces, which has recently been accepted to #NeurIPS2025 !!!

Project: lisaweijler.github.io/openhype-pro...

Are you tired of projecting 2D features into 3D for general problem solving? Try our self-supervised model, which generates semantic off-the-shelf features natively in 3D! #CVPR2025

Project: phermosilla.github.io/msm/
Arxiv: arxiv.org/abs/2504.06719
Github: github.com/phermosilla/...

Check out our latest #CVPR2025 work on masked scene modeling, which learns semantically relevant features by pre-training natively in 3D! 🚀

Love scholar-inbox! I used to rely on Twitter alone to keep up with research, hoping to fully transition to Bluesky + Scholar-Inbox now since the personalized paper recommendations are better than just following some of the X paper accounts.

Agree, pretty sure that's what's happening. Same for other "unknown" object classes such a trike or a quad which also don't have a 3D model and are shown as motorcycle or so.