RegimeVGGT: Layer-Wise Spatially Preserving Redundancy Removal for Visual Geometry Grounded Transformer
cs.CV, cs.RO
Submitted: 2026-06-16
Updated: 2026-11-03
Terminology
Sources
- FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning
- Vision Transformers Need Registers
- Video Token Merging for Long-form Video Understanding
- S-VGGT: Structure-Aware Subscene Decomposition for Scalable 3D Foundation Models
- Not All Patches are What You Need: Expediting Vision Transformers via Token Reorganizations
- DINOv2: Learning Robust Visual Features without Supervision
- FastVGGT: Training-Free Acceleration of Visual Geometry Transformer
- LiteVGGT: Boosting Vanilla VGGT via Geometry-aware Cached Token Merging
- AVGGT: Rethinking Global Attention for Accelerating VGGT
- Block-Sparse Global Attention for Efficient Multi-View Geometry Transformers
- HTTM: Head-wise Temporal Token Merging for Faster VGGT
- DyMU: Dynamic Merging and Virtual Unmerging for Efficient VLMs
- SpargeAttention: Accurate and Training-free Sparse Attention Accelerating Any Model Inference
- Streaming 4D Visual Geometry Transformer
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs