Abstract
Feed-forward 3D vision models such as VGGT have achieved remarkable progress, unifying camera estimation and dense scene reconstruction in a single pass. However, their quadratic global attention makes long image sequences expensive, while existing sparse methods may favor highly attended yet value-redundant regions. To address these limitations, we introduce VASC, a training-free sparse attention method combining value-aware block selection and execution-aware cross-layer memory. Our value-aware block selection integrates pooled query--key relevance with neighboring value contrast, reducing redundancy while preserving query-relevant and distinctive content. Cross-layer memory tracks unserved demand across layers and updates this state according to actual execution, enabling previously underserved blocks to compete under a fixed computation budget. Experiments on 7Scenes and NeuralRGB-D with VGGT and $π^3$ demonstrate improved pose estimation and reconstruction quality compared with FasterVGGT, together with up to $2.29\times$ faster inference than dense VGGT. Code is available at https://github.com/kosakayamahoo-design/VASC.
Keywords
Subject
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Wu, J., Kong, F., Chen, L., Zhang, S., & Zhang, Y. (2026). VASC: Value-Aware Sparse Attention with Cross-Layer Memory for Efficient 3D Reconstruction. https://omanscience.com/en/articles/vasc-value-aware-sparse-attention-with-cross-layer-memory-for-efficient-3d-reconstruction
MLA 9
Wu, Junyi, et al. "VASC: Value-Aware Sparse Attention with Cross-Layer Memory for Efficient 3D Reconstruction." https://omanscience.com/en/articles/vasc-value-aware-sparse-attention-with-cross-layer-memory-for-efficient-3d-reconstruction.
Chicago (author–date)
Wu, Junyi, Fanqing Kong, Leyang Chen, Shaoqiu Zhang, and Yulun Zhang. 2026. "VASC: Value-Aware Sparse Attention with Cross-Layer Memory for Efficient 3D Reconstruction." https://omanscience.com/en/articles/vasc-value-aware-sparse-attention-with-cross-layer-memory-for-efficient-3d-reconstruction.
Harvard
Wu, J., Kong, F., Chen, L., Zhang, S. and Zhang, Y. (2026) 'VASC: Value-Aware Sparse Attention with Cross-Layer Memory for Efficient 3D Reconstruction', Available at: https://omanscience.com/en/articles/vasc-value-aware-sparse-attention-with-cross-layer-memory-for-efficient-3d-reconstruction.
Vancouver
Wu J, Kong F, Chen L, Zhang S, Zhang Y. VASC: Value-Aware Sparse Attention with Cross-Layer Memory for Efficient 3D Reconstruction. https://omanscience.com/en/articles/vasc-value-aware-sparse-attention-with-cross-layer-memory-for-efficient-3d-reconstruction
IEEE
J. Wu, F. Kong, L. Chen, S. Zhang, and Y. Zhang, "VASC: Value-Aware Sparse Attention with Cross-Layer Memory for Efficient 3D Reconstruction," https://omanscience.com/en/articles/vasc-value-aware-sparse-attention-with-cross-layer-memory-for-efficient-3d-reconstruction.