🔗Page: rover-xingyu.github.io/TTT3R
📄Paper: arxiv.org/abs/2509.26645
💻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

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.

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!😀

🔗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

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.

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.

💡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.