Seeing Before Answering: Training-Free Visual Layer Profiling for Vision-Language Models

LLaVA-style Vision-Language Models (VLMs) pass visual tokens from a fixed late layer of the vision backbone, typically the penultimate one, to the language model. We first show that this hidden convention is fragile: across 2 VLMs and 7 image and video benchmarks, the default layer is sub-optimal in 13 of 14 model-task pairs, and the best layer shifts with both task and visual backbone. Finding that layer by exhaustive layer-wise inference is prohibitively expensive, and no better fixed default exists. We therefore ask whether layer usefulness can instead be predicted from representation geometry. We study matrix-based entropy, introduced for unimodal layer analysis, which we compute over sample-level visual embeddings as Visual Dataset Entropy (VDE); and Gromov-Wasserstein (GW) distance, introduced for encoder-level VLM model selection, which we repurpose as a layer-wise visual--language alignment signal. Transferring these to LLaVA-based models is not obvious a priori: the vision tower is frozen while the multimodal projector is trained, so we profile both sides of the projector. We find that VDE transfers, and GW does not. Computed from 100 unlabeled task samples without downstream inference, pre-projector VDE tracks layer-wise accuracy and its top-ranked layers cover the oracle best layer on every task for the SigLIP-based LLaVA-Video, while giving region-level guidance for the CLIP-based Video-LLaVA. Post-projector profiles show that the projector reshapes visual geometry but does not erase the performance-relevant trend, leaving $\mathrm{VDE}_{\mathrm{pre}}$ the stronger signal. GW instead flattens after projection and is best read as an alignment diagnostic rather than a selector. VDE thus offers an interpretable, training-free policy that narrows the visual-layer search to a handful of candidates for limited downstream verification.

OvDSGG: End-to-End Open-Vocabulary Dynamic Scene Graph Generation

Dynamic scene graphs (DSGs) capture spatio-temporal interactions across videos as $\langle$subject, predicate, object$\rangle$ triplets, and underpin downstream tasks such as video captioning, video question answering, and action analysis. However, end-to-end dynamic scene graph generation (DSGG) methods are closed-set: they recognize only objects and predicates from a fixed training vocabulary and struggle with the long-tailed distribution of rare concepts, severely limiting their real-world applicability. Existing open-vocabulary models typically inherit pretrained large language models, resulting in multi-stage training and inference with substantial cost. We introduce OvDSGG, the first end-to-end framework for open-vocabulary DSGG. OvDSGG builds on top of an open-vocabulary Spatial Backbone and a Temporal Backbone; we further propose a Triplet Feature Extraction Module that bridges them, and a Visual-Language Alignment Module that preserves open-vocabulary recognition by learning an adaptive decision boundary in the joint visual-language feature space, without expensive knowledge distillation in existing methods. We further introduce a rigorous open-vocabulary DSGG benchmark adapted from Action Genome, with disjoint Base/Novel splits for both objects and predicates. OvDSGG significantly outperforms open-vocabulary baselines across all metrics, with zero-shot Recall@$K$ scores 10.0--20.4 percentage point higher than the next-best baseline, while on closed-set DSGG remaining competitive with state-of-the-art models. Code and benchmark are publicly available at https://github.com/jhelsby/OvDSGG/.

PUF: Plug-and-Play Uncertainty-Aware Fusion for Online 3D Scene Graph Generation

Online 3D scene graph generation builds a persistent, structured representation of a scene by incrementally fusing 2D observations into a global 3D graph. Existing online methods treat this fusion as a fully deterministic pipeline, where we identify three sources of uncertainty that are overlooked: observation, 2D model, and 3D representation. We propose PUF: a Plug-and-play, Uncertainty-aware, and training-free Fusion framework. Scene graph node association is reformulated as a probabilistic likelihood over semantic and spatial factors, replacing binary accept/reject gates. Dirichlet evidence accumulation distributes class and relationship evidence across plausible candidates proportional to association likelihood. An optional class-conditional prior completes edges for sparsely or never co-observed object pairs. We instantiate PUF with both a 3D Gaussian and a 3D voxel backend and observe consistent improvements, demonstrating its ability to generalize across different representations. Experiments on the 3DSSG and ReplicaSSG benchmarks show that our method substantially outperforms existing approaches while maintaining real-time latency. These results establish uncertainty-aware fusion as a principled and effective paradigm for online 3D scene understanding. The source code is publicly available at https://github.com/yyyyangyi/PUF.

UNIVERSE: Unified Video Action Models for Autonomous Driving with Flexible Mask-Modulated Modality Generation

World Action Models (WAMs) have shown strong potential for improving action generalization in autonomous driving by using future video prediction as dense supervision for scene dynamics and temporal causality. However, it remains unclear which architecture better transfers video-modeling benefits to trajectory generation. Existing cascaded or dual-DiT designs separate video imagination from action prediction, weakening the transfer of video-learned world dynamics to the trajectory branch: the action model may still overfit dataset-specific driving priors, while the video model only indirectly regularizes planning. We propose UNIVERSE, a unified video-action model built upon a single mask-modulated Diffusion Transformer. By co-training future video latents and ego-trajectory tokens within shared generative parameters, UNIVERSE allows dense video supervision to directly shape trajectory denoising, leading to stronger cross-domain action generalization. To ensure causal validity and efficient deployment, we introduce a Modality-Decoupling Visibility Mask, which shares historical context across modalities while blocking mutual attention between future video and trajectory tokens. This prevents future-target leakage and enables trajectory-only inference by removing future-video denoising at test time, achieving a $4.3\times$ speedup over joint video-action rollout while maintaining comparable planning accuracy. The same model also supports video-only and joint video-action rollouts. Experiments show that UNIVERSE achieves 91.0 PDMS on NAVSIM (vs. 89.6 for the Two-DiT variant), and demonstrates strong zero-shot transfer to nuScenes and Bench2Drive without fine-tuning, while ablations confirm the importance of single-DiT unification, video co-training, and mask-based modality decoupling.

Semantic-Driven Scale and Spatial Selection for Efficient Cross-Modal Alignment in Referring Remote Sensing Image Segmentation

Referring Remote Sensing Image Segmentation (RRSIS) seeks to localize and segment the target object or region specified by a natural language expression in a remote sensing image. While existing RRSIS models have benefited from large-scale foundation models, they predominantly rely on full fine-tuning. These approaches are computationally intensive and may weaken the generalization ability of pre-trained models, as extensive fine-tuning on significantly smaller downstream datasets can distort the well-structured feature representations learned during large-scale pre-training. Although Parameter-Efficient Tuning (PET) offers a potential alternative, existing PET frameworks primarily focus on single-modal optimization, failing to capture the complex cross-modal dependencies required for multimodal reasoning, while simultaneously struggling to bridge the substantial domain gap between natural scenes and aerial imagery. To address these limitations, we propose a novel framework, Semantic-driven Scale and Spatial Selection for Efficient Cross-modal Alignment (S4ECA), which enables effective and efficient cross-modal interaction through parameter-efficient adaptation. Specifically, we design a dual-encoder adapter architecture. The textual adapter employs learnable queries to distill highly semantic language proxies from word-level embeddings, facilitating early grounding. Simultaneously, the visual adapter refines hierarchical feature representations through a multi-scale dense extractor, followed by a language-guided scale and spatial selection mechanism that dynamically emphasizes relevant visual contexts, ensuring precise cross-modal alignment. By updating only 2.4% of the backbone parameters, our proposed model achieves state-of-the-art performance on the RRSIS-D and RefSegRS datasets, demonstrating superior efficiency and precision in complex aerial scenarios.

DriveVA: Video Action Models are Zero-Shot Drivers

Generalization is a central challenge in autonomous driving, as real-world deployment requires robust performance under unseen scenarios, sensor domains, and environmental conditions. Recent world-model-based planning methods have shown strong capabilities in scene understanding and multi-modal future prediction, yet their generalization across datasets and sensor configurations remains limited. In addition, their loosely coupled planning paradigm often leads to poor video-trajectory consistency during visual imagination. To overcome these limitations, we propose DriveVA, a novel autonomous driving world model that jointly decodes future visual forecasts and action sequences in a shared latent generative process. DriveVA inherits rich priors on motion dynamics and physical plausibility from well-pretrained large-scale video generation models to capture continuous spatiotemporal evolution and causal interaction patterns. To this end, DriveVA employs a DiT-based decoder to jointly predict future action sequences (trajectories) and videos, enabling tighter alignment between planning and scene evolution. We also introduce a video continuation strategy to strengthen long-duration rollout consistency. DriveVA achieves an impressive closed-loop performance of 90.9 PDM score on the challenge NAVSIM. Extensive experiments also demonstrate the zero-shot capability and cross-domain generalization of DriveVA, which reduces average L2 error and collision rate by 78.9% and 83.3% on nuScenes and 52.5% and 52.4% on the Bench2drive built on CARLA v2 compared with the state-of-the-art world-model-based planner.

RegFormer++: An Efficient Large-Scale 3D LiDAR Point Registration Network with Projection-Aware 2D Transformer

Although point cloud registration has achieved remarkable advances in object-level and indoor scenes, large-scale LiDAR registration methods has been rarely explored before. Challenges mainly arise from the huge point scale, complex point distribution, and numerous outliers within outdoor LiDAR scans. In addition, most existing registration works generally adopt a two-stage paradigm: They first find correspondences by extracting discriminative local descriptors and then leverage robust estimators (e.g. RANSAC) to filter outliers, which are highly dependent on well-designed descriptors and post-processing choices. To address these problems, we propose a novel end-to-end differential transformer network, termed RegFormer++, for large-scale point cloud alignment without requiring any further post-processing. Specifically, a hierarchical projection-aware 2D transformer with linear complexity is proposed to project raw LiDAR points onto a cylindrical surface and extract global point features, which can improve resilience to outliers due to long-range dependencies. Because we fill original 3D coordinates into 2D projected positions, our designed transformer can benefit from both high efficiency in 2D processing and accuracy from 3D geometric information. Furthermore, to effectively reduce wrong point matching, a Bijective Association Transformer (BAT) is designed, combining both cross attention and all-to-all point gathering. To improve training stability and robustness, a feature-transformed optimal transport module is also designed for regressing the final pose transformation. Extensive experiments on KITTI, NuScenes, and Argoverse datasets demonstrate that our model achieves state-of-the-art performance in terms of both accuracy and efficiency.

Query-Guided Spatial-Temporal-Frequency Interaction for Music Audio-Visual Question Answering

Audio--Visual Question Answering (AVQA) is a challenging multimodal task that requires jointly reasoning over audio, visual, and textual information in a given video to answer natural language questions. Inspired by recent advances in Video QA, many existing AVQA approaches primarily focus on visual information processing, leveraging pre-trained models to extract object-level and motion-level representations. However, in those methods, the audio input is primarily treated as complementary to video analysis, and the textual question information contributes minimally to audio--visual understanding, as it is typically integrated only in the final stages of reasoning. To address these limitations, we propose a novel Query-guided Spatial--Temporal--Frequency (QSTar) interaction method, which effectively incorporates question-guided clues and exploits the distinctive frequency-domain characteristics of audio signals, alongside spatial and temporal perception, to enhance audio--visual understanding. Furthermore, we introduce a Query Context Reasoning (QCR) block inspired by prompting, which guides the model to focus more precisely on semantically relevant audio and visual features. Extensive experiments conducted on several AVQA benchmarks demonstrate the effectiveness of our proposed method, achieving significant performance improvements over existing Audio QA, Visual QA, Video QA, and AVQA approaches. The code and pretrained models will be released after publication.

4DSTR: Advancing Generative 4D Gaussians with Spatial-Temporal Rectification for High-Quality and Consistent 4D Generation

Remarkable advances in recent 2D image and 3D shape generation have induced a significant focus on dynamic 4D content generation. However, previous 4D generation methods commonly struggle to maintain spatial-temporal consistency and adapt poorly to rapid temporal variations, due to the lack of effective spatial-temporal modeling. To address these problems, we propose a novel 4D generation network called 4DSTR, which modulates generative 4D Gaussian Splatting with spatial-temporal rectification. Specifically, temporal correlation across generated 4D sequences is designed to rectify deformable scales and rotations and guarantee temporal consistency. Furthermore, an adaptive spatial densification and pruning strategy is proposed to address significant temporal variations by dynamically adding or deleting Gaussian points with the awareness of their pre-frame movements. Extensive experiments demonstrate that our 4DSTR achieves state-of-the-art performance in video-to-4D generation, excelling in reconstruction quality, spatial-temporal consistency, and adaptation to rapid temporal movements.

Scale-wise Bidirectional Alignment Network for Referring Remote Sensing Image Segmentation

The goal of referring remote sensing image segmentation (RRSIS) is to extract specific pixel-level regions within an aerial image via a natural language expression. Recent advancements, particularly Transformer-based fusion designs, have demonstrated remarkable progress in this domain. However, existing methods primarily focus on refining visual features using language-aware guidance during the cross-modal fusion stage, neglecting the complementary vision-to-language flow. This limitation often leads to irrelevant or suboptimal representations. In addition, the diverse spatial scales of ground objects in aerial images pose significant challenges to the visual perception capabilities of existing models when conditioned on textual inputs. In this paper, we propose an innovative framework called Scale-wise Bidirectional Alignment Network (SBANet) to address these challenges for RRSIS. Specifically, we design a Bidirectional Alignment Module (BAM) with learnable query tokens to selectively and effectively represent visual and linguistic features, emphasizing regions associated with key tokens. BAM is further enhanced with a dynamic feature selection block, designed to provide both macro- and micro-level visual features, preserving global context and local details to facilitate more effective cross-modal interaction. Furthermore, SBANet incorporates a text-conditioned channel and spatial aggregator to bridge the gap between the encoder and decoder, enhancing cross-scale information exchange in complex aerial scenarios. Extensive experiments demonstrate that our proposed method achieves superior performance in comparison to previous state-of-the-art methods on the RRSIS-D and RefSegRS datasets, both quantitatively and qualitatively. The code will be released after publication.