Vision-language models, speech and cross-modal systems that read, look and listen in the same forward pass.
arXiv:2609.00059v1 Announce Type: cross Abstract: Materials property prediction remains difficult in low-data settings, where many target properties are supported by only a limited number of labeled...
The paper introduces X‑SG$^2$S, a feed‑forward framework that embeds 1D to 3D watermarks into 3D Gaussian Splatting (3DGS) scenes without altering the original rendering pipeline. It splits watermark messages into patches, uses a self‑adaptive gate to choose injection points, and an XD injection head to embed multi‑modal messages into sorted 3DGS points. A learnable gate and XD‑extraction heads recover the hidden messages, achieving robust watermarking with minimal interference to scene fidelity.
SlideBank is a training‑free framework that turns each whole‑slide image into a persistent, concept‑indexed evidence bank. It performs coarse‑to‑fine exploration to locate informative regions and multi‑scale views, converts them into explicit morphological observations, and anchors pathology signals to the supporting patches and slide coordinates. During inference, questions are routed to relevant signals and evidence scales, and a confidence‑based cross‑level consensus integrates global, regional, and patch evidence, achieving high accuracy on WSI‑VQA and SlideBench‑BCNB while enabling consistent re‑phrasing and reduced inference cost.
The paper investigates whether audio‑language models capture paralinguistic cues beyond spoken content. Using the Expresso dataset and four open‑source models, the authors trace how speaking style information is encoded in the late layers of the audio encoder but is degraded before reaching the final output. They find that some models are content‑driven while others are acoustic‑driven, revealing a gap between what is encoded and what is utilized in current audio‑language models.
arXiv:2609.00377v1 Announce Type: cross Abstract: We present FoldingAgent, an agentic framework for inferring explicit parametric folding programs directly from origami demonstration videos. Our fram...
ADGNet introduces an Asymmetric Dual-text Guided Network for infrared small target detection, addressing challenges of pixel-level methods and multimodal approaches that lack regional guidance. It employs an Asymmetric Dual-text Prompt (ADP) with an abstract target prompt and a detailed background prompt, and an Asymmetric Dual-Branch Interaction (ADBI) module to guide visual features separately, followed by an Adaptive Feature Aggregation (AFA) module for dynamic fusion. The authors also create an Asymmetric Image-Text Infrared (AITIR) dataset with asymmetric text annotations for three public datasets, and show that ADGNet outperforms 21 state‑of‑the‑art methods.
arXiv:2609.00443v1 Announce Type: cross Abstract: Language models learn about grammatical number primarily from co-occurrence, and show frequency effects as a result---sometimes taken to indicate tha...
The paper investigates a failure mode in multimodal large language models called multimodal contextual sycophancy, where external text can override conflicting image evidence. A diagnostic set of 998 cases independently varies visual evidence, commonsense priors, and external text to probe when this failure occurs. Experiments across six models show that a System‑2 Visual Arbitration (S2VA) approach, which withholds text from the visual witness, significantly improves performance over direct witness reports, with the best information boundary varying by model and context source.
arXiv:2510.01030v2 Announce Type: replace Abstract: The human ability to translate diverse perceptual and linguistic inputs into structured behavior has been thought to rest on learning robust repres...
arXiv:2609.00551v1 Announce Type: cross Abstract: Multimodal memory offers a scalable interface for long-video question answering, but existing methods often retrieve captions, frames, transcripts, s...
Instance-Guided Report Anchoring (IGRA) is a model-agnostic module that links each abnormality instance in a chest CT to the corresponding finding in a radiology report during training, while discarding text components at inference. By reformulating free-text grounding as multi-label volumetric segmentation, IGRA allows all abnormality categories to be predicted in a single image-only forward pass. The method improves Dice scores by 22.5% over the strongest image-only baseline and matches state‑of‑the‑art performance on single-finding subsets, with consistent gains across multiple 3D segmentation backbones and datasets.
arXiv:2609.00802v1 Announce Type: new Abstract: Multi-party interaction is a central setting for human communication and a necessary target for human-agent interaction systems that must participate i...
The paper introduces DP-TabImage, a framework that synthesizes paired table–image data while preserving differential privacy. It factorizes the joint distribution into a private probabilistic graphical model for tables and a table‑conditioned diffusion model for images, using DP‑SGD and a privacy‑preserving warm‑up strategy. Experiments on three real datasets demonstrate a strong balance between table fidelity, image fidelity, and cross‑modal alignment.
The paper introduces a new Continual Audio‑Visual Segmentation (CAVS) task that enables continuous segmentation of new classes guided by audio. It identifies two key challenges—multi‑modal semantic drift and co‑occurrence confusion—and proposes a Collision‑based Multi‑modal Rehearsal (CMR) framework with Multi‑modal Sample Selection (MSS) and Collision‑based Sample Rehearsal (CSR) strategies to address them. Experiments on three audio‑visual incremental scenarios show that CMR outperforms single‑modal continual learning methods.
arXiv:2609.00505v1 Announce Type: new Abstract: Video-text models adapted from image-text architectures (e.g., CLIP) frequently exhibit temporal blindness, the inability to perceive fundamental cues...
Vision‑language models used to gauge urban change from repeated street‑level images exhibit limited reliability at single locations. In a study of 4,648 image pairs from 435 Google Street View points across five U.S. cities, re‑photographing the same street altered perception scores by an average of 0.80 points—about two‑thirds of the difference between distinct streets—while repeated model calls added negligible variation. Although image re‑encoding, prompt order, and various image statistics contributed modestly, a small systematic drift (~0.1 points) persisted and grew with time between captures, suggesting minor unrecorded physical changes. Controlled experiments revealed that varying camera and image properties can shift scores, and that camera geometry alone caused a model to falsely report change in 45% of identical scenes; normalising to a common virtual camera reduced this to 7.5%. Despite these individual‑point unreliabilities, aggregating many paired observations recovers a clear redevelopment signal, indicating that such models are dependable at large scales but not for single‑location assessments.
EmbodiedSkills is a unified framework that treats each skill decision as an execution proposal, checking prerequisites and verifying outcomes during long‑horizon vision‑language‑action tasks. It connects high‑level skill selection, bounded low‑level VLA execution, and post‑action verification through a fixed executable‑skill interface, enabling easy replacement of low‑level policies and recording of structured trajectories for supervision and adaptation. Instantiated with Qwen3‑VL and OpenPI/pi0.5 on RoboTwin 2.0 and LIBERO, the framework achieves high success rates (86.20% and 97.40% respectively) and demonstrates effective memory‑dependent task performance.
arXiv:2609.01224v1 Announce Type: new Abstract: Visual token pruning reduces the inference overhead of multimodal large language models (MLLMs) by retaining only a subset of visual tokens. Existing m...
Qwen-Drive-1.0 is a vision‑language foundation model tailored for autonomous driving that builds on a pretrained VLM architecture. It incorporates a bird’s‑eye‑view perception head for 3D object detection, semantic occupancy prediction, and BEV map segmentation, and a Planning Expert that generates future ego trajectories from shared representations. Experiments show strong 3D perception, driving scene understanding, and competitive motion‑planning performance while largely preserving general vision‑language capabilities.
arXiv:2603.29852v2 Announce Type: replace-cross Abstract: We introduce VectorGym, a comprehensive benchmark suite for Scalable Vector Graphics (SVG) that spans generation from text and sketches, comp...