arXiv subjects

Jungong Han

Publications and source records attributed to Jungong Han.

At least 55 records · Page 3Linked to original sources

arXiv2026

Point Linguist Model: Segment Any Object via Bridged Large 3D-Language Model

3D object segmentation with Large Language Models (LLMs) has become a prevailing paradigm due to its broad semantics, task flexibility, and strong generalization. However, this paradigm is hindered by representation misalignment: LLMs process high-level semantic tokens, whereas 3D point clouds convey only dense geometric structures. In prior methods, misalignment limits both input and output. At the input stage, dense point patches require heavy pre-alignment, weakening object-level semantics and confusing similar distractors. At the output stage, predictions depend only on dense features without explicit geometric cues, leading to a loss of fine-grained accuracy. To address these limitations, we present the Point Linguist Model (PLM), a general framework that bridges the representation gap between LLMs and dense 3D point clouds without requiring large-scale pre-alignment between 3D-text or 3D-images. Specifically, we introduce Object-centric Discriminative Representation (OcDR), which learns object-centric tokens that capture target semantics and scene relations under a hard negative-aware training objective. This mitigates the misalignment between LLM tokens and 3D points, enhances resilience to distractors, and facilitates semantic-level reasoning within LLMs. For accurate segmentation, we introduce the Geometric Reactivation Decoder (GRD), which predicts masks by combining OcDR tokens carrying LLM-inferred geometry with corresponding dense features, preserving comprehensive dense features throughout the pipeline. Extensive experiments show that PLM achieves significant improvements of +7.3 mIoU on ScanNetv2 and +6.0 mIoU on Multi3DRefer for 3D referring segmentation, with consistent gains across 7 benchmarks spanning 4 different tasks, demonstrating the effectiveness of comprehensive object-centric reasoning for robust 3D understanding.

cs.CV↗

arXiv2026

UniAttn: Reducing Inference Costs via Softmax Unification for Post-Training LLMs

Post-training is essential for adapting Large Language Models (LLMs) to real-world applications. Deploying post-trained models faces significant challenges due to substantial memory overhead and noticeable inference latency. Existing work has identified significant redundancies in LLMs and proposed efficient architectures, namely intra-layer KV sharing and cross-layer KV sharing. However, these methods still result in high inference time overhead, remaining suboptimal for post-training pre-trained LLMs. In this paper, we identify that the \texttt{Softmax} operation is a primary bottleneck for LLM inference and discover that it is actually highly redundant during post-training. We propose Softmax \textbf{Uni}fication in \textbf{Att}e\textbf{n}tion (\textbf{UniAttn}), a novel post-training method that unifies Softmax activations across transformer blocks to reduce LLM inference costs. Additionally, UniAttn adopts a linear projection to compensate for the errors induced by Softmax unification. Experiments show that UniAttn matches the performance of standard post-training while significantly reducing inference costs, outperforming existing efficient architectures during post-training.

cs.CL↗

arXiv2025

SAM-Body4D: Training-Free 4D Human Body Mesh Recovery from Videos

Human Mesh Recovery (HMR) aims to reconstruct 3D human pose and shape from 2D observations and is fundamental to human-centric understanding in real-world scenarios. While recent image-based HMR methods such as SAM 3D Body achieve strong robustness on in-the-wild images, they rely on per-frame inference when applied to videos, leading to temporal inconsistency and degraded performance under occlusions. We address these issues without extra training by leveraging the inherent human continuity in videos. We propose SAM-Body4D, a training-free framework for temporally consistent and occlusion-robust HMR from videos. We first generate identity-consistent masklets using a promptable video segmentation model, then refine them with an Occlusion-Aware module to recover missing regions. The refined masklets guide SAM 3D Body to produce consistent full-body mesh trajectories, while a padding-based parallel strategy enables efficient multi-human inference. Experimental results demonstrate that SAM-Body4D achieves improved temporal stability and robustness in challenging in-the-wild videos, without any retraining. Our code and demo are available at: https://github.com/gaomingqi/sam-body4d.

cs.CV↗

arXiv2025

Tracking and Segmenting Anything in Any Modality

Tracking and segmentation play essential roles in video understanding, providing basic positional information and temporal association of objects within video sequences. Despite their shared objective, existing approaches often tackle these tasks using specialized architectures or modality-specific parameters, limiting their generalization and scalability. Recent efforts have attempted to unify multiple tracking and segmentation subtasks from the perspectives of any modality input or multi-task inference. However, these approaches tend to overlook two critical challenges: the distributional gap across different modalities and the feature representation gap across tasks. These issues hinder effective cross-task and cross-modal knowledge sharing, ultimately constraining the development of a true generalist model. To address these limitations, we propose a universal tracking and segmentation framework named SATA, which unifies a broad spectrum of tracking and segmentation subtasks with any modality input. Specifically, a Decoupled Mixture-of-Expert (DeMoE) mechanism is presented to decouple the unified representation learning task into the modeling process of cross-modal shared knowledge and specific information, thus enabling the model to maintain flexibility while enhancing generalization. Additionally, we introduce a Task-aware Multi-object Tracking (TaMOT) pipeline to unify all the task outputs as a unified set of instances with calibrated ID information, thereby alleviating the degradation of task-specific knowledge during multi-task training. SATA demonstrates superior performance on 18 challenging tracking and segmentation benchmarks, offering a novel perspective for more generalizable video understanding.

cs.CV↗

arXiv2025

PruneHal: Reducing Hallucinations in Multi-modal Large Language Models through Adaptive KV Cache Pruning

While multi-modal large language models (MLLMs) have made significant progress in recent years, the issue of hallucinations remains a major challenge. To mitigate this phenomenon, existing solutions either introduce additional data for further training or incorporate external or internal information during inference. However, these approaches inevitably introduce extra computational costs. In this paper, we observe that hallucinations in MLLMs are strongly associated with insufficient attention allocated to visual tokens. In particular, the presence of redundant visual tokens disperses the model's attention, preventing it from focusing on the most informative ones. As a result, critical visual cues are often under-attended, which in turn exacerbates the occurrence of hallucinations. Building on this observation, we propose \textbf{PruneHal}, a training-free, simple yet effective method that leverages adaptive KV cache pruning to enhance the model's focus on critical visual information, thereby mitigating hallucinations. To the best of our knowledge, we are the first to apply token pruning for hallucination mitigation in MLLMs. Notably, our method don't require additional training and incurs nearly no extra inference cost. Moreover, PruneHal is model-agnostic and can be seamlessly integrated with different decoding strategies, including those specifically designed for hallucination mitigation. We evaluate PruneHal on several widely used hallucination evaluation benchmarks using four mainstream MLLMs, achieving robust and outstanding results that highlight the effectiveness and superiority of our method. Our code will be publicly available.

cs.CV↗

arXiv2025

CAIT: Triple-Win Compression towards High Accuracy, Fast Inference, and Favorable Transferability For ViTs

Vision Transformers (ViTs) have emerged as state-of-the-art models for various vision tasks recently. However, their heavy computation costs remain daunting for resource-limited devices. To address this, researchers have dedicated themselves to compressing redundant information in ViTs for acceleration. However, existing approaches generally sparsely drop redundant image tokens by token pruning or brutally remove channels by channel pruning, leading to a sub-optimal balance between model performance and inference speed. Moreover, they struggle when transferring compressed models to downstream vision tasks that require the spatial structure of images, such as semantic segmentation. To tackle these issues, we propose CAIT, a joint \underline{c}ompression method for ViTs that achieves a harmonious blend of high \underline{a}ccuracy, fast \underline{i}nference speed, and favorable \underline{t}ransferability to downstream tasks. Specifically, we introduce an asymmetric token merging (ATME) strategy to effectively integrate neighboring tokens. It can successfully compress redundant token information while preserving the spatial structure of images. On top of it, we further design a consistent dynamic channel pruning (CDCP) strategy to dynamically prune unimportant channels in ViTs. Thanks to CDCP, insignificant channels in multi-head self-attention modules of ViTs can be pruned uniformly, significantly enhancing the model compression. Extensive experiments on multiple benchmark datasets show that our proposed method can achieve state-of-the-art performance across various ViTs.

cs.CV↗

arXiv2025

PrefixKV: Adaptive Prefix KV Cache is What Vision Instruction-Following Models Need for Efficient Generation

Recently, large vision-language models (LVLMs) have rapidly gained popularity for their strong generation and reasoning capabilities given diverse multimodal inputs. However, these models incur significant computational and memory overhead during inference, which greatly hinders the efficient deployment in practical scenarios. The extensive key-value (KV) cache, necessitated by the lengthy input and output sequences, notably contributes to the high inference cost. Based on this, recent works have investigated ways to reduce the KV cache size for higher efficiency. Although effective, they generally overlook the distinct importance distributions of KV vectors across layers and maintain the same cache size for each layer during the next token prediction. This results in the significant contextual information loss for certain layers, leading to notable performance decline. To address this, we present PrefixKV, where "Prefix" means the top-ranked KV based on importance rather than position in the original sequence. It reframes the challenge of determining KV cache sizes for all layers into the task of searching for the optimal global prefix configuration. With an adaptive layer-wise KV retention recipe based on binary search, the maximum contextual information can thus be preserved in each layer, facilitating the generation. Extensive experiments demonstrate that our method achieves the state-of-the-art performance compared with others. It exhibits superior inference efficiency and generation quality trade-offs, showing promising potential for practical applications. Code is available at https://github.com/THU-MIG/PrefixKV.

cs.CV↗

arXiv2025

YOLOE: Real-Time Seeing Anything

Object detection and segmentation are widely employed in computer vision applications, yet conventional models like YOLO series, while efficient and accurate, are limited by predefined categories, hindering adaptability in open scenarios. Recent open-set methods leverage text prompts, visual cues, or prompt-free paradigm to overcome this, but often compromise between performance and efficiency due to high computational demands or deployment complexity. In this work, we introduce YOLOE, which integrates detection and segmentation across diverse open prompt mechanisms within a single highly efficient model, achieving real-time seeing anything. For text prompts, we propose Re-parameterizable Region-Text Alignment (RepRTA) strategy. It refines pretrained textual embeddings via a re-parameterizable lightweight auxiliary network and enhances visual-textual alignment with zero inference and transferring overhead. For visual prompts, we present Semantic-Activated Visual Prompt Encoder (SAVPE). It employs decoupled semantic and activation branches to bring improved visual embedding and accuracy with minimal complexity. For prompt-free scenario, we introduce Lazy Region-Prompt Contrast (LRPC) strategy. It utilizes a built-in large vocabulary and specialized embedding to identify all objects, avoiding costly language model dependency. Extensive experiments show YOLOE's exceptional zero-shot performance and transferability with high inference efficiency and low training cost. Notably, on LVIS, with 3$\times$ less training cost and 1.4$\times$ inference speedup, YOLOE-v8-S surpasses YOLO-Worldv2-S by 3.5 AP. When transferring to COCO, YOLOE-v8-L achieves 0.6 AP$^b$ and 0.4 AP$^m$ gains over closed-set YOLOv8-L with nearly 4$\times$ less training time. Code and models are available at https://github.com/THU-MIG/yoloe.

cs.CV↗

arXiv2025

LSVOS 2025 Challenge Report: Recent Advances in Complex Video Object Segmentation

This report presents an overview of the 7th Large-scale Video Object Segmentation (LSVOS) Challenge held in conjunction with ICCV 2025. Besides the two traditional tracks of LSVOS that jointly target robustness in realistic video scenarios: Classic VOS (VOS), and Referring VOS (RVOS), the 2025 edition features a newly introduced track, Complex VOS (MOSEv2). Building upon prior insights, MOSEv2 substantially increases difficulty, introducing more challenging but realistic scenarios including denser small objects, frequent disappear/reappear events, severe occlusions, adverse weather and lighting, etc., pushing long-term consistency and generalization beyond curated benchmarks. The challenge retains standard ${J}$, $F$, and ${J\&F}$ metrics for VOS and RVOS, while MOSEv2 adopts ${J\&\dot{F}}$ as the primary ranking metric to better evaluate objects across scales and disappearance cases. We summarize datasets and protocols, highlight top-performing solutions, and distill emerging trends, such as the growing role of LLM/MLLM components and memory-aware propagation, aiming to chart future directions for resilient, language-aware video segmentation in the wild.

cs.CV↗

arXiv2025

From Gaze to Insight: Bridging Human Visual Attention and Vision Language Model Explanation for Weakly-Supervised Medical Image Segmentation

Medical image segmentation remains challenging due to the high cost of pixel-level annotations for training. In the context of weak supervision, clinician gaze data captures regions of diagnostic interest; however, its sparsity limits its use for segmentation. In contrast, vision-language models (VLMs) provide semantic context through textual descriptions but lack the explanation precision required. Recognizing that neither source alone suffices, we propose a teacher-student framework that integrates both gaze and language supervision, leveraging their complementary strengths. Our key insight is that gaze data indicates where clinicians focus during diagnosis, while VLMs explain why those regions are significant. To implement this, the teacher model first learns from gaze points enhanced by VLM-generated descriptions of lesion morphology, establishing a foundation for guiding the student model. The teacher then directs the student through three strategies: (1) Multi-scale feature alignment to fuse visual cues with textual semantics; (2) Confidence-weighted consistency constraints to focus on reliable predictions; (3) Adaptive masking to limit error propagation in uncertain areas. Experiments on the Kvasir-SEG, NCI-ISBI, and ISIC datasets show that our method achieves Dice scores of 80.78%, 80.53%, and 84.22%, respectively-improving 3-5% over gaze baselines without increasing the annotation burden. By preserving correlations among predictions, gaze data, and lesion descriptions, our framework also maintains clinical interpretability. This work illustrates how integrating human visual attention with AI-generated semantic context can effectively overcome the limitations of individual weak supervision signals, thereby advancing the development of deployable, annotation-efficient medical AI systems. Code is available at: https://github.com/jingkunchen/FGI.

cs.CV↗

arXiv2025

The 1st Solution for MOSEv2 Challenge 2025: Long-term and Concept-aware Video Segmentation via SeC

This technical report explores the MOSEv2 track of the LSVOS Challenge, which targets complex semi-supervised video object segmentation. By analysing and adapting SeC, an enhanced SAM-2 framework, we conduct a detailed study of its long-term memory and concept-aware memory, showing that long-term memory preserves temporal continuity under occlusion and reappearance, while concept-aware memory supplies semantic priors that suppress distractors; together, these traits directly benefit several MOSEv2's core challenges. Our solution achieves a JF score of 39.89% on the test set, ranking 1st in the MOSEv2 track of the LSVOS Challenge.

cs.CV↗

arXiv2025

Temporal Scaling Law for Large Language Models

Recently, Large Language Models (LLMs) have been widely adopted in a wide range of tasks, leading to increasing attention towards the research on how scaling LLMs affects their performance. Existing works, termed Scaling Laws, have discovered that the final test loss of LLMs scales as power-laws with model size, computational budget, and dataset size. However, the temporal change of the test loss of an LLM throughout its pre-training process remains unexplored, though it is valuable in many aspects, such as selecting better hyperparameters \textit{directly} on the target LLM. In this paper, we propose the novel concept of Temporal Scaling Law, studying how the test loss of an LLM evolves as the training steps scale up. In contrast to modeling the test loss as a whole in a coarse-grained manner, we break it down and dive into the fine-grained test loss of each token position, and further develop a dynamic hyperbolic-law. Afterwards, we derive the much more precise temporal scaling law by studying the temporal patterns of the parameters in the dynamic hyperbolic-law. Results on both in-distribution (ID) and out-of-distribution (OOD) validation datasets demonstrate that our temporal scaling law accurately predicts the test loss of LLMs across training steps. Our temporal scaling law has broad practical applications. First, it enables direct and efficient hyperparameter selection on the target LLM, such as data mixture proportions. Secondly, viewing the LLM pre-training dynamics from the token position granularity provides some insights to enhance the understanding of LLM pre-training.

cs.CL↗

arXiv2025

DSMoE: Matrix-Partitioned Experts with Dynamic Routing for Computation-Efficient Dense LLMs

As large language models continue to scale, computational costs and resource consumption have emerged as significant challenges. While existing sparsification methods like pruning reduce computational overhead, they risk losing model knowledge through parameter removal. This paper proposes DSMoE (Dynamic Sparse Mixture-of-Experts), a novel approach that achieves sparsification by partitioning pre-trained FFN layers into computational blocks. We implement adaptive expert routing using sigmoid activation and straight-through estimators, enabling tokens to flexibly access different aspects of model knowledge based on input complexity. Additionally, we introduce a sparsity loss term to balance performance and computational efficiency. Extensive experiments on LLaMA models demonstrate that under equivalent computational constraints, DSMoE achieves superior performance compared to existing pruning and MoE approaches across language modeling and downstream tasks, particularly excelling in generation tasks. Analysis reveals that DSMoE learns distinctive layerwise activation patterns, providing new insights for future MoE architecture design.

cs.CL↗

arXiv2025

Advancing Reliable Test-Time Adaptation of Vision-Language Models under Visual Variations

Vision-language models (VLMs) exhibit remarkable zero-shot capabilities but struggle with distribution shifts in downstream tasks when labeled data is unavailable, which has motivated the development of Test-Time Adaptation (TTA) to improve VLMs' performance during inference without annotations. Among various TTA approaches, cache-based methods show promise by preserving historical knowledge from low-entropy samples in a dynamic cache and fostering efficient adaptation. However, these methods face two critical reliability challenges: (1) entropy often becomes unreliable under distribution shifts, causing error accumulation in the cache and degradation in adaptation performance; (2) the final predictions may be unreliable due to inflexible decision boundaries that fail to accommodate large downstream shifts. To address these challenges, we propose a Reliable Test-time Adaptation (ReTA) method that integrates two complementary strategies to enhance reliability from two perspectives. First, to mitigate the unreliability of entropy as a sample selection criterion for cache construction, we introduce Consistency-aware Entropy Reweighting (CER), which incorporates consistency constraints to weight entropy during cache updating. While conventional approaches rely solely on low entropy for cache prioritization and risk introducing noise, our method leverages predictive consistency to maintain a high-quality cache and facilitate more robust adaptation. Second, we present Diversity-driven Distribution Calibration (DDC), which models class-wise text embeddings as multivariate Gaussian distributions, enabling adaptive decision boundaries for more accurate predictions across visually diverse content. Extensive experiments demonstrate that ReTA consistently outperforms state-of-the-art methods, particularly under real-world distribution shifts. Code: https://github.com/Evelyn1ywliang/ReTA.

cs.CV↗

arXiv2025

Unlocking the Potential of Diffusion Priors in Blind Face Restoration

Although diffusion prior is rising as a powerful solution for blind face restoration (BFR), the inherent gap between the vanilla diffusion model and BFR settings hinders its seamless adaptation. The gap mainly stems from the discrepancy between 1) high-quality (HQ) and low-quality (LQ) images and 2) synthesized and real-world images. The vanilla diffusion model is trained on images with no or less degradations, whereas BFR handles moderately to severely degraded images. Additionally, LQ images used for training are synthesized by a naive degradation model with limited degradation patterns, which fails to simulate complex and unknown degradations in real-world scenarios. In this work, we use a unified network FLIPNET that switches between two modes to resolve specific gaps. In Restoration mode, the model gradually integrates BFR-oriented features and face embeddings from LQ images to achieve authentic and faithful face restoration. In Degradation mode, the model synthesizes real-world like degraded images based on the knowledge learned from real-world degradation datasets. Extensive evaluations on benchmark datasets show that our model 1) outperforms previous diffusion prior based BFR methods in terms of authenticity and fidelity, and 2) outperforms the naive degradation model in modeling the real-world degradations.

cs.CV↗

arXiv2025

SimMLM: A Simple Framework for Multi-modal Learning with Missing Modality

In this paper, we propose SimMLM, a simple yet powerful framework for multimodal learning with missing modalities. Unlike existing approaches that rely on sophisticated network architectures or complex data imputation techniques, SimMLM provides a generic and effective solution that can adapt to various missing modality scenarios with improved accuracy and robustness. Specifically, SimMLM consists of a generic Dynamic Mixture of Modality Experts (DMoME) architecture, featuring a dynamic, learnable gating mechanism that automatically adjusts each modality's contribution in both full and partial modality settings. A key innovation of SimMLM is the proposed More vs. Fewer (MoFe) ranking loss, which ensures that task accuracy improves or remains stable as more modalities are made available. This aligns the model with an intuitive principle: removing one or more modalities should not increase accuracy. We validate SimMLM on multimodal medical image segmentation (BraTS 2018) and multimodal classification (UPMC Food-101, avMNIST) tasks, where it consistently surpasses competitive methods, demonstrating superior accuracy, interpretability, robustness, and reliability across both complete and missing modality scenarios at test time.

cs.CV↗

arXiv2025

Neutralizing Token Aggregation via Information Augmentation for Efficient Test-Time Adaptation

Test-Time Adaptation (TTA) has emerged as an effective solution for adapting Vision Transformers (ViT) to distribution shifts without additional training data. However, existing TTA methods often incur substantial computational overhead, limiting their applicability in resource-constrained real-world scenarios. To reduce inference cost, plug-and-play token aggregation methods merge redundant tokens in ViTs to reduce total processed tokens. Albeit efficient, it suffers from significant performance degradation when directly integrated with existing TTA methods. We formalize this problem as Efficient Test-Time Adaptation (ETTA), seeking to preserve the adaptation capability of TTA while reducing inference latency. In this paper, we first provide a theoretical analysis from a novel mutual information perspective, showing that token aggregation inherently leads to information loss, which cannot be fully mitigated by conventional norm-tuning-based TTA methods. Guided by this insight, we propose to \textbf{N}eutralize Token \textbf{A}ggregation \textbf{v}ia \textbf{I}nformation \textbf{A}ugmentation (\textbf{NAVIA}). Specifically, we directly augment the [CLS] token embedding and incorporate adaptive biases into the [CLS] token in shallow layers of ViTs. We theoretically demonstrate that these augmentations, when optimized via entropy minimization, recover the information lost due to token aggregation. Extensive experiments across various out-of-distribution benchmarks demonstrate that NAVIA significantly outperforms state-of-the-art methods by over 2.5\%, while achieving an inference latency reduction of more than 20\%, effectively addressing the ETTA challenge.

cs.CV↗

arXiv2025

SAIGFormer: A Spatially-Adaptive Illumination-Guided Network for Low-Light Image Enhancement

Recent Transformer-based low-light enhancement methods have made promising progress in recovering global illumination. However, they still struggle with non-uniform lighting scenarios, such as backlit and shadow, appearing as over-exposure or inadequate brightness restoration. To address this challenge, we present a Spatially-Adaptive Illumination-Guided Transformer (SAIGFormer) framework that enables accurate illumination restoration. Specifically, we propose a dynamic integral image representation to model the spatially-varying illumination, and further construct a novel Spatially-Adaptive Integral Illumination Estimator ($\text{SAI}^2\text{E}$). Moreover, we introduce an Illumination-Guided Multi-head Self-Attention (IG-MSA) mechanism, which leverages the illumination to calibrate the lightness-relevant features toward visual-pleased illumination enhancement. Extensive experiments on five standard low-light datasets and a cross-domain benchmark (LOL-Blur) demonstrate that our SAIGFormer significantly outperforms state-of-the-art methods in both quantitative and qualitative metrics. In particular, our method achieves superior performance in non-uniform illumination enhancement while exhibiting strong generalization capabilities across multiple datasets. Code is available at https://github.com/LHTcode/SAIGFormer.git.

cs.CV↗