AVGGT: Rethinking Global Attention for Accelerating VGGT
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914380455608320 |
|---|---|
| author | Sun, Xianbing Zhu, Zhikai Lou, Zhengyu Yang, Bo Tang, Jinyang Zhang, Liqing Wang, He Zhang, Jianfu |
| author_facet | Sun, Xianbing Zhu, Zhikai Lou, Zhengyu Yang, Bo Tang, Jinyang Zhang, Liqing Wang, He Zhang, Jianfu |
| contents | Models such as VGGT and $π^3$ have shown strong multi-view 3D performance, but their heavy reliance on global self-attention results in high computational cost. Existing sparse-attention variants offer partial speedups, yet lack a systematic analysis of how global attention contributes to multi-view reasoning. In this paper, we first conduct an in-depth investigation of the global attention modules in VGGT and $π^3$ to better understand their roles. Our analysis reveals a clear division of roles in the alternating global-frame architecture: early global layers do not form meaningful correspondences, middle layers perform cross-view alignment, and last layers provide only minor refinements. Guided by these findings, we propose a training-free two-step acceleration scheme: (1) converting early global layers into frame attention, and (2) subsampling global attention by subsampling K/V over patch tokens with diagonal preservation and a mean-fill component. We instantiate this strategy on VGGT and $π^3$ and evaluate across standard pose and point-map benchmarks. Our method achieves substantial inference acceleration across different context lengths, yielding about $2\times$ speedup at 100 frames, $4$--$5\times$ at 300 frames, and $8$--$10\times$ at 800 frames, while matching or slightly improving the accuracy of the original models and remaining robust in extremely dense multi-view settings where prior sparse-attention baselines fail. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_02541 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | AVGGT: Rethinking Global Attention for Accelerating VGGT Sun, Xianbing Zhu, Zhikai Lou, Zhengyu Yang, Bo Tang, Jinyang Zhang, Liqing Wang, He Zhang, Jianfu Computer Vision and Pattern Recognition Models such as VGGT and $π^3$ have shown strong multi-view 3D performance, but their heavy reliance on global self-attention results in high computational cost. Existing sparse-attention variants offer partial speedups, yet lack a systematic analysis of how global attention contributes to multi-view reasoning. In this paper, we first conduct an in-depth investigation of the global attention modules in VGGT and $π^3$ to better understand their roles. Our analysis reveals a clear division of roles in the alternating global-frame architecture: early global layers do not form meaningful correspondences, middle layers perform cross-view alignment, and last layers provide only minor refinements. Guided by these findings, we propose a training-free two-step acceleration scheme: (1) converting early global layers into frame attention, and (2) subsampling global attention by subsampling K/V over patch tokens with diagonal preservation and a mean-fill component. We instantiate this strategy on VGGT and $π^3$ and evaluate across standard pose and point-map benchmarks. Our method achieves substantial inference acceleration across different context lengths, yielding about $2\times$ speedup at 100 frames, $4$--$5\times$ at 300 frames, and $8$--$10\times$ at 800 frames, while matching or slightly improving the accuracy of the original models and remaining robust in extremely dense multi-view settings where prior sparse-attention baselines fail. |
| title | AVGGT: Rethinking Global Attention for Accelerating VGGT |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2512.02541 |