The paper introduces Parabolic Positional Encoding (PaPE), a parabola‑based method for encoding positions of vision tokens in attention‑based models. PaPE is designed to respect vision‑specific properties such as translation and rotation invariance, distance decay, directionality, and context awareness. Experiments on ImageNet‑1K and eight other datasets across four modalities show that PaPE extrapolates well—improving performance by up to 10.5 points over the next best encoding—and generally matches or surpasses baseline encodings.
arXiv:2606. 31585v1 Announce Type: cross Abstract: The remarkable scalability of Transformers has expanded their application to 3D computer vision, where camera-aware positional encoding is crucial for providing spatial cues in multi-view geometry.
arXiv:2609.21212v1 Announce Type: cross Abstract: Learned navigation policies typically consume observations as a temporally ordered history, with positional encodings tying each observation to when...
arXiv:2606. 14757v1 Announce Type: cross Abstract: Though Vision Transformers (ViTs) have become the dominant backbone in many computer vision tasks, due to permutation equivariance, their attention mechanism lacks explicit spatial inductive biases.
arXiv:2605. 19398v3 Announce Type: replace-cross Abstract: Image-to-video models often generate videos that remain overly static, compared to text-to-video models.
arXiv:2601. 15275v3 Announce Type: replace-cross Abstract: We study positional encodings for multi-view transformers that process tokens from a set of posed input images, and seek a mechanism that encodes patches uniquely, allows SE(3)-invariant attention with multi-frequency similarity, and can adapt to the geometry of the underlying 3D scene.
arXiv:2606. 00124v1 Announce Type: cross Abstract: Positional embeddings (PEs) in Vision Transformers (ViTs) are known to impact performance and robustness, but their role in shaping internal spatial representations is not well understood.
The paper introduces AttWarp, a lightweight technique that uses a multimodal large language model’s cross‑modal attention to perform rectilinear warping of input images at test time. By reallocating spatial resolution toward query‑relevant regions without altering model weights or architecture, AttWarp preserves global context while making small objects and subtle relationships easier for the model to read. Experiments on five benchmarks and four MLLMs show consistent accuracy gains, improved compositional reasoning, and reduced hallucinations compared to baseline image‑manipulation methods.
arXiv:2609.01252v1 Announce Type: new Abstract: In camera-controlled video generation, geometry-aware positional encodings condition tokens on camera extrinsics and per-token viewing rays. Existing s...
arXiv:2609.38109v1 Announce Type: cross Abstract: The attention operation is naively position invariant. However, positional information is fundamental to natural language, and therefore a variety of...
LearnPruner rethinks token pruning in vision‑language models by addressing shortcomings of attention mechanisms in vision encoders and large language models. The method uses a two‑stage approach: first a learnable pruning module removes redundant vision tokens after the vision encoder, then only task‑relevant tokens are kept in the LLM’s middle layer. Experiments show that LearnPruner retains about 95% of performance while using only 5.5% of vision tokens, achieving a 3.2× speed‑up.
The paper introduces Adaptive Visual Token Pruning (AVTP), a training‑free framework that dynamically selects pruning layers and ratios for large vision‑language models (LVLMs) when processing multiple image sequences. By analyzing visual attention distributions across different LVLM architectures, AVTP adapts token retention to image importance, enabling efficient inference without relying on attention‑based computations incompatible with FlashAttention. Experiments show significant speedups—up to 2× for Qwen3VL‑8B—while preserving or even improving accuracy on multi‑image benchmarks.
DK‑VLM introduces Dual‑Kernel Attention, where a visual token’s attention is computed from both its input position and its geometric position derived from 3D points and camera poses. A semantic kernel preserves pretrained attention while a geometric kernel modulates attention via point displacements and relative view transformations, with a learned gate fusing the two. The model also employs Reference‑Frame Composition supervision to update its reference frame across successive judgments, achieving top scores on six spatial benchmarks and strong performance on VSI‑Bench and ReVSI‑32.