Fuxin Li

@CSProfKGD I surely think you would know for sure😀

@fuxinli2 I do 🤓

Did you know that Gaussian splatting always assumed that none of the Gaussians overlap, which is definitely not a correct assumption? Our CVPR 2026 (findings) effort Softmax-GS improves over the non-overlapping assumption in Gaussian splatting, and make multiple Gaussians compete in regions where they overlap using a softmax function, which greatly improves rendering performance when less Gaussians are used, as well as rendering consistency from different viewpoints. Especially, check out our experiments fitting multiple geometric shapes with only 4 Gaussians! arthurhero.github.io/projects/smgs/

A bit worried that not a lot of final ratings have been submitted and there is only a little over a day left. Reviewers please note there is this one last task from you by May 20th!

@tdietterich @arxiv Thanks so much Tom for the service to arXiv and trying to keep the community sane and professional!

Attention @arxiv authors: Our Code of Conduct states that by signing your name as an author of a paper, each author takes full responsibility for all its contents, irrespective of how the contents were generated. 1/

@CSProfKGD I don't support having all authors in the reviewer pool. In CVPR 2025 we only asked qualified first authors and authors who submitted multiple papers, and I still wished we asked less people than that.

#KostasThoughts: Perhaps we’ve gone too far with requiring ALL authors to be in the reviewer pool in computer vision conferences and instead we should consider a per paper nomination system 🤔

Still time to submit for non-archival papers and participate in the NSS challenge! Please consider this very interesting and practical application of 3D vision!

@_jessethomason_ Congratulations Jesse!

This fall semester, I will be moving my GLAMOR lab (glamor.rocks) to the School of Interactive Computing (IC) at the Georgia Institute of Technology!

Long-LRM++ is our newest high-resolution feedforward scene reconstruction, to appear in CVPR 2026 (findings). With feature splatting, we got significantly more clarity — small font texts can now be clearly rendered in a few seconds!

Deeply grateful for the NVIDIA Academic Grant Program to support us with precious compute for advancing the research on 3D foundational models for embodied AI! #NVIDIAGrant @NVIDIAAIDev

Where Computer Vision meets the Built Environment. 📐✨ The #CV4AEC Workshop is coming to @CVPR in Denver! We’re moving beyond static vision to explore how multimodal data models the evolution of our physical world. Speakers, paper & challenge deadlines 👇 cv4aec.github.io

Our new work exploring whole body controller in long range planning tasks on humanoids. Key finding: the policy for the skills can stay unchanged but improving the whole body controller is key to long-horizon successes.

Prospective faculty would be joining an excellent cohort of robotics faculty, students and robots at Oregon State!

I'll be at NeuRIPS and happy to provide more information. DM please.

Are you passionate about connecting 3D perception / world modeling with learning-based control for humanoid mobile manipulation in the real world? The Dynamic Robotics and AI Lab (DRAIL) mime.engineering.oregonstate.edu/research/drl/i… at Oregon State University is recruiting 2–3 PhD students to help build next-generation humanoid systems that perceive the 3D world and interact with it. At DRAIL, we work at the cutting edge of robotics control, deep learning, computer vision, and reinforcement learning, enabling legged robots to both move and manipulate with agility. Our research has earned multiple best paper awards across top venues in computer vision (CVPR), machine learning (ICLR), and robotics (ICRA). We also set world records in bipedal robot locomotion such as Fastest 100m dash by a bipedal robot (2022) youtube.com/watch?v=64hKiu… and 📷 First bipedal robot to complete a 5K (2021) youtube.com/watch?v=VOBnIb… 📷 You will also be powered by state-of-the-art GPUs that are going to come to the Huang complex in 2026, thanks to the generous donation of our alumnae Mr. Jen-Hsun Huang and Mrs. Lori Mills Huang. huangcomplex.oregonstate.edu/inside-complex If you’re excited about enabling robots to perceive and act in complex real-world environments, we’d love to hear from you. 📷 Please apply to Oregon State University’s Robotics (deadline 12/15) graduate.oregonstate.edu/programs/3250/… or AI (deadline 12/10) graduate program. graduate.oregonstate.edu/programs/3075/…

@CSProfKGD I am now on the side of "as long as the review is sound and correct scientifically (which the reviewer is supposed to guarantee) we should be happy with it".

@fuxinli2 This is another problem where some authors are automatically accusing reviewers of delegating reviews to LLMs when they don’t like the content of the reviews 😒

#KostasThoughts: Please stop using AI tools to write reviews. I want your authentic thoughts, even if they contain spelling, grammar, & factual mistakes. An LLM should not influence your judgment. The purpose of multiple reviewers is to provide diverse views and error checking.