arXiv subjects

Xiaolei Chen

Publications and source records attributed to Xiaolei Chen.

10 recordsLinked to original sources

arXiv2026

CRISP: Pre-LLM Yet Text-Driven Visual Token Pruning for Efficient LVLM Inference

Large Vision-Language Models (LVLMs) typically require processing hundreds to thousands of visual tokens, leading to substantial inference overhead. Existing visual token pruning methods either operate before the LLM using text-agnostic heuristics or prune inside the LLM at the cost of efficiency and noisy cross-modal attention. To address these limitations, we propose CRISP, a pre-LLM yet text-driven visual token pruning framework that preserves both instruction-relevant evidence and essential scene context. CRISP works in a two-stage pipeline: Stage 1 first identifies text-aligned visual tokens, and Stage 2 enhances contextual completeness through semantic diversity. Extensive experiments on LLaVA-1.5 and LLaVA-NeXT demonstrate that CRISP achieves superior performance retention under aggressive pruning ratios, maintaining up to 99.5% accuracy while reducing inference cost and latency by more than 2 times. CRISP serves as a practical solution for efficient LVLM inference, especially in resource-constrained scenarios.

cs.CV↗

arXiv2026

The Curious Case of PHL 1811: Heavy Obscuration Versus Intrinsic X-ray Weakness

We present a systematic X-ray analysis of the narrow-line Type 1 quasar PHL 1811, which has long been regarded as the prototype of intrinsically X-ray weak quasars. A critical breakthrough came with the first detection of a bright X-ray flare from this source by the Einstein Probe (EP) in 2024. We utilize archival X-ray observations spanning 2001-2024, including the post-flare EP and Swift data. We confirm that PHL 1811 shows X-ray weakness factors $f_{\rm weak} \approx 23$-179 across all epochs before 2024. The 2024 EP flare marks the first detection of an X-ray nominal state with $f_{\rm weak} \approx 0.63$, followed by a rapid flux decline. We identify three key observational signatures that strongly support heavy obscuration: (1) a significant hard X-ray excess above $\approx5$ keV in the 2015 XMM-Newton spectrum; (2) relatively flat spectral shapes in two Swift observations; and (3) transitions between X-ray nominal and multiple X-ray weak states without corresponding optical/infrared variability, consistent with expectations from obscuration by a clumpy dust-free absorber. Fitting with a partial-covering obscuration model reproduces all multi-epoch spectra well. The observed steep spectra are dominated by a small leaked/scattered fraction of the intrinsic continuum, and variability is driven by changes in the leakage fraction and column density. Our results strongly favor the scenario where PHL 1811 is obscured by a radiatively driven accretion-disk wind from super-Eddington accretion, unifying PHL 1811 with the broader population of super-Eddington accreting AGNs under a single obscuration framework.

astro-ph.HE↗

arXiv2026

ResPrune: Text-Conditioned Subspace Reconstruction for Visual Token Pruning in Large Vision-Language Models

Large Vision-Language Models (LVLMs) rely on dense visual tokens to capture fine-grained visual information, but processing all these tokens incurs substantial computational and memory overhead during inference. To address this issue, we propose ResPrune, a training-free visual token pruning framework that enables efficient LVLM inference by selecting a compact yet informative subset of visual tokens. ResPrune formulates visual token pruning as a subspace reconstruction problem and employs a greedy subspace expansion strategy guided by residual energy, allowing it to preserve the geometric structure of the original visual token space. To further incorporate cross modal alignment, the selection process is conditioned on textual relevance, encouraging the retention of tokens that are both informative and instruction-relevant. The proposed method is lightweight and model-agnostic, and can be seamlessly integrated into existing LVLM pipelines without retraining or architectural modifications. Extensive experiments on multiple LVLM backbones, including LLaVA-1.5, LLaVA-NeXT, and Qwen2.5-VL, demonstrate that ResPrune consistently outperforms existing pruning approaches across a wide range of benchmarks, while achieving effective reductions in computation, memory consumption, and inference latency.

cs.LG↗

arXiv2026

DynamicVGGT: Learning Dynamic Point Maps for 4D Scene Reconstruction in Autonomous Driving

Dynamic scene reconstruction in autonomous driving remains a fundamental challenge due to significant temporal variations, moving objects, and complex scene dynamics. Existing feed-forward 3D models have demonstrated strong performance in static reconstruction but still struggle to capture dynamic motion. To address these limitations, we propose DynamicVGGT, a unified feed-forward framework that extends VGGT from static 3D perception to dynamic 4D reconstruction. Our goal is to model point motion within feed-forward 3D models in a dynamic and temporally coherent manner. To this end, we jointly predict the current and future point maps within a shared reference coordinate system, allowing the model to implicitly learn dynamic point representations through temporal correspondence. To efficiently capture temporal dependencies, we introduce a Motion-aware Temporal Attention (MTA) module that learns motion continuity. Furthermore, we design a Dynamic 3D Gaussian Splatting Head that explicitly models point motion by predicting Gaussian velocities using learnable motion tokens under scene flow supervision. It refines dynamic geometry through continuous 3D Gaussian optimization. Extensive experiments on autonomous driving datasets demonstrate that DynamicVGGT significantly outperforms existing methods in reconstruction accuracy, achieving robust feed-forward 4D dynamic scene reconstruction under complex driving scenarios.

cs.CV↗

arXiv2025

HERO: Rethinking Visual Token Early Dropping in High-Resolution Large Vision-Language Models

By cropping high-resolution images into local tiles and encoding them independently, High-Resolution Large Vision-Language Models (HR-LVLMs) have demonstrated remarkable fine-grained visual understanding capabilities. However, this divide-and-conquer paradigm significantly increases the number of visual tokens, resulting in substantial computational and memory overhead. To better understand and address this challenge, we empirically investigate visual token utilization in HR-LVLMs and uncover three key findings: (1) the local tiles have varying importance, jointly determined by visual saliency and task relevance; (2) the CLS token in CLIP-based vision encoders exhibits a two-stage attention pattern across layers, with each stage attending to different types of visual tokens; (3) the visual tokens emphasized at different stages encode information at varying levels of granularity, playing complementary roles within LVLMs. Building on these insights, we propose HERO, a High-resolution visual token early dropping framework that integrates content-adaptive token budget allocation with function-aware token selection. By accurately estimating tile-level importance and selectively retaining visual tokens with complementary roles, HERO achieves superior efficiency-accuracy trade-offs across diverse benchmarks and model scales, all in a training-free manner. This study provides both empirical insights and practical solutions toward efficient inference in HR-LVLMs.

cs.CV↗

arXiv2025

Pyramid Token Pruning for High-Resolution Large Vision-Language Models via Region, Token, and Instruction-Guided Importance

Large Vision-Language Models (LVLMs) have recently demonstrated strong multimodal understanding, yet their fine-grained visual perception is often constrained by low input resolutions. A common remedy is to partition high-resolution images into multiple sub-images for separate encoding, but this approach drastically inflates the number of visual tokens and introduces prohibitive inference overhead. To overcome this challenge, we propose Pyramid Token Pruning (PTP), a training-free strategy that hierarchically integrates bottom-up visual saliency at both region and token levels with top-down instruction-guided relevance. Inspired by human visual cognition, PTP selectively preserves more tokens from salient regions while further emphasizing those most relevant to task instructions. Extensive experiments on 13 diverse benchmarks show that PTP substantially reduces computational cost, memory usage, and inference latency, with negligible performance degradation.

cs.CV↗

arXiv2025

Global Semantic-Guided Sub-image Feature Weight Allocation in High-Resolution Large Vision-Language Models

As the demand for high-resolution image processing in Large Vision-Language Models (LVLMs) grows, sub-image partitioning has become a popular approach for mitigating visual information loss associated with fixed-resolution processing. However, existing partitioning methods uniformly process sub-images, resulting in suboptimal image understanding. In this work, we reveal that the sub-images with higher semantic relevance to the entire image encapsulate richer visual information for preserving the model's visual understanding ability. Therefore, we propose the Global Semantic-guided Weight Allocator (GSWA) module, which dynamically allocates weights to sub-images based on their relative information density, emulating human visual attention mechanisms. This approach enables the model to focus on more informative regions, overcoming the limitations of uniform treatment. We integrate GSWA into the InternVL2-2B framework to create SleighVL, a lightweight yet high-performing model. Extensive experiments demonstrate that SleighVL outperforms models with comparable parameters and remains competitive with larger models. Our work provides a promising direction for more efficient and contextually aware high-resolution image processing in LVLMs, advancing multimodal system development.

cs.CV↗

arXiv2025

Int2Planner: An Intention-based Multi-modal Motion Planner for Integrated Prediction and Planning

Motion planning is a critical module in autonomous driving, with the primary challenge of uncertainty caused by interactions with other participants. As most previous methods treat prediction and planning as separate tasks, it is difficult to model these interactions. Furthermore, since the route path navigates ego vehicles to a predefined destination, it provides relatively stable intentions for ego vehicles and helps constrain uncertainty. On this basis, we construct Int2Planner, an \textbf{Int}ention-based \textbf{Int}egrated motion \textbf{Planner} achieves multi-modal planning and prediction. Instead of static intention points, Int2Planner utilizes route intention points for ego vehicles and generates corresponding planning trajectories for each intention point to facilitate multi-modal planning. The experiments on the private dataset and the public nuPlan benchmark show the effectiveness of route intention points, and Int2Planner achieves state-of-the-art performance. We also deploy it in real-world vehicles and have conducted autonomous driving for hundreds of kilometers in urban areas. It further verifies that Int2Planner can continuously interact with the traffic environment. Code will be avaliable at https://github.com/cxlz/Int2Planner.

cs.RO↗

arXiv2025

Instruction-Guided Fusion of Multi-Layer Visual Features in Large Vision-Language Models

Large Vision-Language Models (LVLMs) have achieved remarkable success in a wide range of multimodal tasks by integrating pre-trained vision encoders and large language models. However, current LVLMs primarily rely on visual features extracted from the final layers of the vision encoder, overlooking the complementary information available in shallower layers. While recent approaches have explored the use of multilayer visual features in LVLMs, they tend to be task-agnostic and fail to examine the dependencies of hierarchical visual features on specific tasks. To address these gaps, we systematically investigate the contributions of visual features from different encoder layers using 18 benchmarks spanning 6 task categories. Our findings reveal that multilayer features provide complementary strengths with varying task dependencies, and uniform fusion leads to suboptimal performance. Building on these insights, we propose the instruction-guided vision aggregator, a module that dynamically integrates multi-layer visual features based on textual instructions, without increasing the number of visual tokens. Extensive evaluations demonstrate the superior performance of our method. Additionally, an in-depth analysis of the aggregator's behavior highlights the dominance of mid-to-high-level features in semantic-rich tasks and the critical role of low-level features in fine-grained perception.

cs.CV↗

arXiv2023

A Spatial-Temporal Dual-Mode Mixed Flow Network for Panoramic Video Salient Object Detection

Salient object detection (SOD) in panoramic video is still in the initial exploration stage. The indirect application of 2D video SOD method to the detection of salient objects in panoramic video has many unmet challenges, such as low detection accuracy, high model complexity, and poor generalization performance. To overcome these hurdles, we design an Inter-Layer Attention (ILA) module, an Inter-Layer weight (ILW) module, and a Bi-Modal Attention (BMA) module. Based on these modules, we propose a Spatial-Temporal Dual-Mode Mixed Flow Network (STDMMF-Net) that exploits the spatial flow of panoramic video and the corresponding optical flow for SOD. First, the ILA module calculates the attention between adjacent level features of consecutive frames of panoramic video to improve the accuracy of extracting salient object features from the spatial flow. Then, the ILW module quantifies the salient object information contained in the features of each level to improve the fusion efficiency of the features of each level in the mixed flow. Finally, the BMA module improves the detection accuracy of STDMMF-Net. A large number of subjective and objective experimental results testify that the proposed method demonstrates better detection accuracy than the state-of-the-art (SOTA) methods. Moreover, the comprehensive performance of the proposed method is better in terms of memory required for model inference, testing time, complexity, and generalization performance.

cs.CV↗