iiinova — where your digital self keeps evolving

Xingyu Chen @xingyu-chen.bsky.social · 28/05/2026

Predict pairwise poses with confidence, then assemble the trajectory. The right design doesn’t just follow the scaling curve; it reshapes it.

000

Reposted by Xingyu Chen

Kwang Moo Yi @kmyid.bsky.social · 03/12/2025

Hu, Cheng, Yu et al., "VGGT4D: Mining Motion Cues in Visual Geometry Transformers for 4D Scene Reconstruction" Easi3r-style attention analysis and masking with mask refinement with VGGT. Also discards tokens related to dynamic points.

121

Reposted by Xingyu Chen

Andreas Geiger @andreasgeiger.bsky.social · 10/10/2025

Personal programs for ICCV 2025 are now available at: www.scholar-inbox.com/conference/i...

0246

Xingyu Chen @xingyu-chen.bsky.social · 08/10/2025

Look, 4D foundation models know about humans – and we just read it out!

010

Xingyu Chen @xingyu-chen.bsky.social · 05/10/2025

Glad to be recognized as an outstanding reviewer!

020

Xingyu Chen @xingyu-chen.bsky.social · 01/10/2025

Instead of updating all states uniformly, we incorporate image attention as per-token learning rates. High-confidence matches get larger updates, while low-quality updates are suppressed. This soft gating greatly extends the length generalization beyond the training context.

110

Xingyu Chen @xingyu-chen.bsky.social · 01/10/2025

#VGGT: accurate within short clips, but slow and prone to Out-of-Memory (OOM) #CUT3R: fast with constant memory usage, but forgets. We revisit them from a Test-Time Training (TTT) perspective and propose #TTT3R to get all three: fast, accurate, and OOM-free.

111

Xingyu Chen @xingyu-chen.bsky.social · 01/10/2025

Let's keep revisiting 3D reconstruction!

020

Reposted by Xingyu Chen

Haiwen Huang @haiwen-huang.bsky.social · 22/04/2025

Excited to introduce LoftUp! A strong (than ever) and lightweight feature upsampler for vision encoders that can boost performance on dense prediction tasks by 20%–100%! Easy to plug into models like DINOv2, CLIP, SigLIP — simple design, big gains. Try it out! github.com/andrehuang/l...

0195

Xingyu Chen @xingyu-chen.bsky.social · 15/04/2025

If you're a researcher and haven't tried it yet, please give it a try! It took me a while to adjust, but now it's my favorite tool. You can read, bookmark, organize papers, and get recommendations based on your interests!

010

Reposted by Xingyu Chen

Zhenjun Zhao @ericzzj.bsky.social · 02/04/2025

Easi3R: Estimating Disentangled Motion from DUSt3R Without Training @xingyu-chen.bsky.social, @fanegg.bsky.social, @xiuyuliang.bsky.social, @andreasgeiger.bsky.social, @apchen.bsky.social arxiv.org/abs/2503.24391

162

Reposted by Xingyu Chen

Vision and Graphics Trends @si-cv-graphics.bsky.social · 02/04/2025

𝗘𝗮𝘀𝗶𝟯𝗥: 𝗘𝘀𝘁𝗶𝗺𝗮𝘁𝗶𝗻𝗴 𝗗𝗶𝘀𝗲𝗻𝘁𝗮𝗻𝗴𝗹𝗲𝗱 𝗠𝗼𝘁𝗶𝗼𝗻 𝗳𝗿𝗼𝗺 𝗗𝗨𝗦𝘁𝟯𝗥 𝗪𝗶𝘁𝗵𝗼𝘂𝘁 𝗧𝗿𝗮𝗶𝗻𝗶𝗻𝗴 Xingyu Chen, Yue Chen, Yuliang Xiu ... Anpei Chen arxiv.org/abs/2503.24391 Trending on www.scholar-inbox.com

012

Xingyu Chen @xingyu-chen.bsky.social · 02/04/2025

DUSt3R was never trained to do dynamic segmentation with GT masks, right? It was just trained to regress point maps on 3D datasets—yet dynamic awareness emerged, making DUSt3R a zero-shot 4D estimator!😀

140

Reposted by Xingyu Chen

Andreas Geiger @andreasgeiger.bsky.social · 01/04/2025

I was really surprised when I saw this. Dust3R has learned very well to segment objects without supervision. This knowledge can be extracted post-hoc, enabling accurate 4D reconstruction instantly.

1312

Xingyu Chen @xingyu-chen.bsky.social · 01/04/2025

🔗Page: easi3r.github.io 📄Paper: arxiv.org/abs/2503.24391 💻Code: github.com/Inception3D/... Big thanks to the amazing team! @xingyu-chen.bsky.social, @fanegg.bsky.social, @xiuyuliang.bsky.social, @andreasgeiger.bsky.social, @apchen.bsky.social

030

Xingyu Chen @xingyu-chen.bsky.social · 01/04/2025

With our estimated segmentation masks, we perform a second inference pass by re-weighting the attention, enabling robust 4D reconstruction and even outperforming SOTA methods trained on 4D datasets, with almost no extra cost compared to vanilla DUSt3R.

140

Xingyu Chen @xingyu-chen.bsky.social · 01/04/2025

We propose an attention-guided strategy to decompose dynamic objects from the static background, enabling robust dynamic object segmentation. It outperforms the optical-flow guided segmentation, like MonST3R, and the model trained on dynamic mask labels, like DAS3R.

130

Xingyu Chen @xingyu-chen.bsky.social · 01/04/2025

💡Humans naturally separate ego-motion from object-motion without dynamic labels. We observe that #DUSt3R has implicitly learned a similar mechanism, reflected in its attention layers.

130

Xingyu Chen @xingyu-chen.bsky.social · 01/04/2025

🦣Easi3R: 4D Reconstruction Without Training! Limited 4D datasets? Take it easy. #Easi3R adapts #DUSt3R for 4D reconstruction by disentangling and repurposing its attention maps → make 4D reconstruction easier than ever! 🔗Page: easi3r.github.io

2223

Reposted by Xingyu Chen

Yue Chen @fanegg.bsky.social · 31/03/2025

How much 3D do visual foundation models (VFMs) know? Previous work requires 3D data for probing → expensive to collect! #Feat2GS @cvprconference.bsky.social 2025 - our idea is to read out 3D Gaussains from VFMs features, thus probe 3D with novel view synthesis. 🔗Page: fanegg.github.io/Feat2GS

1247