Iro Armeni

GitHub - GradientSpaces/Rectified-Point-Flow: Rectified Point Flow - Assemble unposed parts into complete objects by learning a point-wise flow model. Rectified Point Flow - Assemble unposed parts into complete objects by learning a point-wise flow model. - GradientSpaces/Rectified-Point-Flow

📄 arxiv.org/abs/2506.05282
💻 github.com/GradientSpac...
🤹 w/ Tao Sun, @liyuanzzz.bsky.social, Shengyu Huang, and Shuran Song

Just dropped: Rectified Point Flow

Can we automate 3D assembly from unposed point clouds without supervision?
Yes—we use a generative model that learns symmetry & part interchangeability entirely from shape

Enabling robotics, AR/VR, & reverse engineering

🌐 rectified-pointflow.github.io

Research scholar program Overview

🌟🏆 Excited to share that I’ve been selected as a recipient of the 2025 Google Research Scholar Program, on Machine Perception! Grateful to Google Research for supporting early-career faculty.

Details and this year’s cohort of exceptional researchers: goo.gle/rs

🤹 w/ the driving force @azadef.bsky.social, ‪@eadeli.bsky.social‬, Federico Tombari, Nassir Navab.

SG2RL Overview The workshop focuses on the topic of scene graphs and graph representation learning for visual perception applications in different domains. Through a series of keynote talks, the audience wi...

🚨Only 1 day left to submit a full paper on (scene) graphs to the 3rd Workshop on #SG2RL!
- Can't make it? Extended abstracts are due on July 7th
- Dangling carrot🥕? We are in 𝑯𝒂𝒘𝒂𝒊'𝒊, @ICCVConference 2025
🌐 sites.google.com/view/sg2rl/

Correction: We’ll be there from 12PM tomorrow.

We are here today till 5:15PM and tomorrow from 10AM.

Creators (alphabetically): @ir0armeni.bsky.social, @mnbucher.bsky.social , @sayandsarkar.bsky.social , Emily Steiner, Tao Sun, @jianhao97.bsky.social , @liyuanzzz.bsky.social

At @cvprconference.bsky.social? Come to the AI Art exhibit to interact with our #GradientCanvas!

It is a community generated #AR painting. Users can claim a region and add their own contribution. No painting skills required, just words!
gradientcanvas.github.io

Gradients Spaces, Stanford Univ.

(4/4)
▶️Gradient Canvas: An AR Community Art Project
🌐https://gradientcanvas.github.io
🤹w/(alphab.): @mnbucher.bsky.social, @sayandsarkar.bsky.social, Emily Steiner, Tao Sun, @jianhao97.bsky.social, @liyuanzzz.bsky.social
📍CVPR AI Art
📅Jun Fri 13th-Sun 15th
Find us and interact with our AR canvas!

(3/4)
▶️ WildGS-SLAM: Monocular Gaussian Splatting SLAM in Dynamic Environments
🌐https://wildgs-slam.github.io
🤹 w/ @jianhao97.bsky.social,Zihan Zhu,Valentin Bieri,@marcpollefeys.bsky.social,@songyoupeng.bsky.social
📅 Poster, Sat June 14th, 10:30-12:30PM
📅 Demo: Sun June 15th, 10:30-12:30PM
📍ExHallD

(2/4)
▶️CrossOver: 3D Scene Cross-Modal Alignment
🌐https://sayands.github.io/crossover/
🤹@sayandsarkar.bsky.social‬, Ondrej Miksik, @marcpollefeys.bsky.social
📅Highlight Poster #346, Fri June 13th, 4-6PM
📍ExHallD
& 4th Workshop on Open-World 3D Scene Underst. w/ Foundation Models, June 12th, Poster

At @cvprconference.bsky.social? Join me for:
(1/4)
▶️ Talk at 5th Workshop on 3D Scene Understanding for Vision, Graphics, and Robotics
📅 Wed June 11th, 11:15-11:45AM
📍 106C

📰 Paper: arxiv.org/abs/2502.15011
▶️ Project Page: sayands.github.io/crossover/
💻 Codebase: github.com/GradientSpaces…

Work w/ Ondrej Miksik, @marcpollefeys.bsky.social, @danielbarath.bsky.social and @ir0armeni.bsky.social ✨

(3/3)

Meet 𝐑𝐞𝐒𝐩𝐚𝐜𝐞, a text-driven 3D scene synthesis+editing method, powered by autoregressive LLMs and spatially grounded representations.

- Object addition = next-token prediction
- RLVR-tuned for fine control
- Voxel loss for geometric nuance

🌐 respace.mnbucher.com
w/ @mnbucher.bsky.social

📜https://arxiv.org/abs/2502.10377
⌨️https://github.com/GradientSpaces/ReStyle3D
🤗https://huggingface.co/gradient-spaces/ReStyle3D

🤹 w/ @liyuanzzz.bsky.social, Shengqu Cai, Shengyu Huang, Gordon Wetzstein, and Naji Khosravan

🎨 AI-powered interior design from video—just bring a Pinterest image.
Meet ReStyle3D #SIGGRAPH25: per-object, multi-view style transfer for real scenes.
No 3D model. No camera poses. Just semantics + diffusion magic.
🚀 Stylize scenes, photorealistically.
🌐 restyle3d.github.io

🚨 Winners of the Nothing Stands Still Challenge 2025 are in!
🥇First Place: Daebeom Kim, Seungjae Lee, Seoyeon Jang, Jei Kong, Prof. Hyun Myung
🥈Second Place: Xuan Ouyang, Linjie Li, Zhengbao Wang, Prof. Siwen Quan, Prof. Jiaqi Yang
Congrats! 🏗️✨
#ICRA2025 #NSS

Nothing Stands Still — Except the Date!
Join us May 19 (Atlanta or Zoom) for the 4th Workshop on Construction Robotics!
Explore 4D perception, automation, & results from the #NothingStandsStill challenge.
Info: construction-robots.github.io
#ConstructionRobotics #4DPerception

🚨 👉 NSS Challenge deadline extended to **May 12th**-- spread the word!

🏗️ Nothings Stands Still Challenge 2025
🌐 www.nothing-stands-still.com/challenge

4th Workshop on Future of Construction, ICRA 2025
May 19th, Atlanta, USA
construction-robots.github.io

#NSS #ICRA2025

🤹 w/ @jianhao97.bsky.social, @zhuzihan2000.bsky.social, Valentin Bieri, Marc Pollefeys, @songyoupeng.bsky.social

🚨 SLAM struggling in dynamic environments? We've been there.

WildGS-SLAM at #CVPR2025, our new monocular RGB SLAM system, tackles dynamic scenes with uncertainty-aware tracking and mapping, resulting to more robust tracking, cleaner maps, and high-quality view synthesis. ⬇️

🌐 wildgs-slam.github.io

🌐 Workshop Details: construction-robots.github.io
🤹 Workshop Organizers: Jingdao Chen, Yong Kwon Cho, Inbae Jeong, Chen Feng, Liangjun Zhang, @mauricefallon.bsky.social, Michael Helmberger, Kristian Morin, Borja García de Soto, Xi Wang, Ci-Jyun Liang, and me.

This is the second iteration of the challenge hosted together with HILTI at the '4th Future of Construction Workshop', #ICRA2025.

🤹 Challenge Organizers:
Tao Sun, Emily Steiner, and me (Gradient Spaces)
Torben Graeber and Michael Helmberger (HILTI Group)

The Nothing Stands Still Challenge is back and due on ⌛ May 7th!
💰 Kept the cash prize but created new data that are extra challenging.
🏗️ Goal: create spatiotemporal maps of drastically changing environments. What a better fit than construction sites?

Visit here:
🌐 hilti-challenge.com

Joint work with @sayandsarkar.bsky.social, Ondrej Miksik, @marcpollefeys.bsky.social, and @danielbarath.bsky.social

How do you align point clouds, CAD, floor plans. images, and text to share scene knowledge?
CrossOver learns a unified, modality-agnostic embedding space—enabling seamless multi-modal scene alignment without semantic labels.

#CVPR2025
🌐sayands.github.io/crossover/