arXiv subjects

Longxiang Tang

Publications and source records attributed to Longxiang Tang.

At least 19 recordsLinked to original sources

arXiv2026

Behind Video Generation: Observations and Insights on Multimodal Understanding from Kling

Video generation has progressed from isolated clips to coherent short-form audiovisual sequences, and the bottleneck is shifting from rendering to semantic control. This report summarizes the observations and insights on multimodal understanding gained from successive generations of Kling. We argue that generation quality is the projection of multimodal understanding through the entire data--conditioning--alignment--evaluation chain, and we examine this chain module by module. Kling-MLLM, a shared multimodal foundation model that unifies image, video, audio, synchronized audio--video, and interleaved inputs, is specialized for each downstream module. Because a DiT is trained on far fewer training samples than a language model, we treat captioning as a semantic contract that maximizes description density, and extend it to reference-grounded captioning that makes reference bindings explicit. Prompt enhancement translates user intent into this caption distribution: basic PE is derived from captions of compelling videos and refined with text-level and generator-in-the-loop reinforcement learning, while agentic PE refines the story and retrieves compelling examples. We show that a reliable text encoder comes from perceptual training rather than from stronger language ability, and that candidates can be screened from text alone before end-to-end training. Reward modeling pairs a general reward model for explainable, localized critiques with a decision reward model for fast multidimensional scoring, treating perceptual bandwidth as a first-class design parameter. Finally, we review task-specific automatic evaluation, focusing on protocols that expose scoring shortcuts and help trace failures to their upstream source. We hope these lessons offer practical guidance for building the understanding stack behind controllable video generation.

cs.CV↗

arXiv2026

Beyond Data Scaling: Representation-Centric Continued Pre-training for Vision-Language-Action Models

Scaling robot data is crucial for building generalist Vision-Language-Action (VLA) models, yet robot trajectories are harder to scale than web-scale image-text data because embodied collection is costly and sparsely covers the physical world. This makes representation quality a central bottleneck: under a fixed robot-data budget, continued pre-training must turn limited trajectories into transferable visual-action knowledge rather than merely fit actions. We propose VLAct, a VLA-oriented VLM backbone trained on broad, heterogeneous, multi-embodiment robot data before task-specific fine-tuning. VLAct preserves the broad VLM prior and encourages shared action semantics across embodiments through VLM-prior preservation, multi-head continuous action co-supervision, and a partially unified cross-embodiment action layout, while allowing task-specific action heads during fine-tuning. Across simulation, real-world, and unseen-embodiment transfer, VLAct consistently improves downstream performance under fixed fine-tuning protocols. On LIBERO-Plus and RoboTwin 2.0, VLAct surpasses industrial VLA systems including ABot-M0 and LingBot-VLA, achieving success rates of 82.6% and 92.5%. On RoboDojo, VLAct ranks sixth among all policies by success rate and outperforms all explicitly designated world-action model (WAM) entries on both metrics. Most notably, on RoboCasa-GR1, an unseen humanoid embodiment, VLAct using only 20% of downstream trajectories outperforms the full-data GR00T-N1.6 baseline. These results are obtained using fully open-source data and only a 16-GPU training setup, showing that representation-centric continued pre-training can deliver highly competitive performance under a modest compute budget and is an important independent axis of VLA progress beyond data scaling.

cs.RO↗

arXiv2026

UnfoldCRF: Structured Mask Refinement with Image-Conditioned Latent Regions

Learned mask refiners improve segmentation accuracy, but it is hard to tell how much of the improvement comes from explicit structure rather than from extra capacity, and whether it holds up when the mask generator or its error distribution changes. UnfoldCRF treats refinement as inference in a conditional random field over pixel labels and latent region variables. Its energy has a corrected unary term, learned local pairwise interactions, and image-conditioned latent-region consistency, with a null state that lets a region with weak label agreement withdraw from the consistency term; inference unrolls damped mean-field updates on this one energy. To isolate the effect of structure, we compare against recurrent black-box refiners that read the same inputs and receive the same parameter budget, stage count, and supervision. On COD10K, UnfoldCRF beats the strongest matched control by 1.0 $F^ω_β$ point, improves all four COD metrics, and lowers the fraction of images made worse from 11.7\% to 8.5\%. Under a train-once protocol over five datasets and several mask sources, the 2.6M-parameter variant gains 4.2 mean $Δ$IoU against 2.0 for its control, and a variant built on frozen DINOv2 features matches the strongest foundation-model refiner with about a seventh of its resident parameters while staying ahead of its own control. On mask generators never seen in training, the gain is 2.0 $F^ω_β$ points against 0.6 for the control. Zeroing individual messages shows where the corrections come from: the pairwise messages mostly fix boundaries, the region messages mostly fix non-boundary errors. Code and supporting materials will be publicly released.

cs.CV↗

arXiv2026

RePlan: Reasoning-guided Region Planning for Complex Instruction-based Image Editing

Instruction-based image editing enables natural-language control over visual modifications, yet existing models falter under Instruction-Visual Complexity (IV-Complexity), where intricate instructions meet cluttered or ambiguous scenes. We introduce RePlan (Region-aligned Planning), a plan-then-execute framework that couples a vision-language planner with a diffusion editor. The planner decomposes instructions via step-by-step reasoning and explicitly grounds them to target regions; the editor then applies changes using a training-free attention-region injection mechanism, enabling precise, parallel multi-region edits without iterative inpainting. To strengthen planning, we apply GRPO-based reinforcement learning using 1K instruction-only examples, yielding substantial gains in reasoning fidelity and format reliability. We further present IV-Edit, a benchmark focused on fine-grained grounding and knowledge-intensive edits. Across IV-Complex settings, RePlan consistently outperforms strong baselines trained on far larger datasets, improving regional precision and overall consistency. Our project page: https://replan-iv-edit.github.io

cs.CV↗

arXiv2026

PRISM: Rethinking Atmospheric Scattering Reconstruction as a Unified Understanding and Restoration Model for Real-world Dehazing

Real-world image dehazing (RID) aims to remove haze-induced degradation from real scenes. This task remains challenging due to non-uniform haze distribution, spatially varying color shifts, and the scarcity of paired real hazy-clean data. In PRISM, we propose Proximal Scattering Atmosphere Reconstruction (PSAR), a physically structured framework that jointly reconstructs the clear scene and scattering variables under the atmospheric scattering model, making the restoration process more interpretable in complex real-world conditions. To bridge the synthetic-to-real gap, we design an online non-uniform haze synthesis pipeline and a Selective Self-Distillation Adaptation (SSDA) scheme for unpaired real-world scenarios, which enables the model to selectively learn from high-quality perceptual targets while leveraging its intrinsic scattering understanding to audit residual haze and guide self-refinement. Experiments on real-world benchmarks demonstrate that PRISM achieves competitive performance on RID tasks.

cs.CV↗

arXiv2026

RIDE: Retinex-Informed Decoupling for Exposing Concealed Objects

Concealed Object Segmentation (COS) encompasses a family of dense-prediction tasks, including camouflaged object detection, polyp segmentation, transparent object detection, and industrial defect inspection, where targets are visually entangled with their surroundings through different physical mechanisms. Existing methods either operate directly on RGB images or employ \emph{heterogeneous} decompositions (\eg, Fourier, wavelet) that redistribute spatial evidence across scale/frequency coefficients, making pixel-aligned cues less direct. We introduce a fundamentally different perspective: \textbf{homogeneous image decomposition} via Retinex theory, which factorizes an image into illumination and reflectance components within the \emph{same} spatial domain. Our key insight is that visual entanglement enforces appearance matching in the composite space, but this does \emph{not} necessitate simultaneous matching in both component spaces, a phenomenon we formalize as the \textbf{Discriminability Gap Theorem}. Crucially, we show that across diverse COS sub-tasks, the underlying physical processes systematically anti-correlate illumination and reflectance differences, yielding theoretical guarantees that Retinex decomposition preserves or strictly improves total foreground--background discriminability across the full physical regime, with anti-correlation maximizing the gain. Building on this, we propose \textbf{RIDE} comprising: (i) a Task-Driven Retinex Decomposition module that learns segmentation-optimal factorizations end-to-end; (ii) a Discriminability Gap Attention mechanism that adaptively exploits where decomposition helps; and (iii) a Camouflage-Breaking Contrastive loss operating in reflectance feature space.

cs.CV↗

arXiv2026

A Vision-Language-Action Model for Adaptive Ultrasound-Guided Needle Insertion and Needle Tracking

Ultrasound (US)-guided needle insertion is a critical yet challenging procedure due to dynamic imaging conditions and difficulties in needle visualization. Many methods have been proposed for automated needle insertion, but they often rely on hand-crafted pipelines with modular controllers, whose performance degrades in challenging cases. In this paper, a Vision-Language-Action (VLA) model is proposed for adaptive and automated US-guided needle insertion and tracking on a robotic ultrasound (RUS) system. This framework provides a unified approach to needle tracking and needle insertion control, enabling real-time, dynamically adaptive adjustment of insertion based on the obtained needle position and environment awareness. To achieve real-time and end-to-end tracking, a Cross-Depth Fusion (CDF) tracking head is proposed, integrating shallow positional and deep semantic features from the large-scale vision backbone. To adapt the pretrained vision backbone for tracking tasks, a Tracking-Conditioning (TraCon) register is introduced for parameter-efficient feature conditioning. After needle tracking, an uncertainty-aware control policy and an asynchronous VLA pipeline are presented for adaptive needle insertion control, ensuring timely decision-making for improved safety and outcomes. Extensive experiments on both needle tracking and insertion show that our method consistently outperforms state-of-the-art trackers and manual operation, achieving higher tracking accuracy, improved insertion success rates, and reduced procedure time, highlighting promising directions for RUS-based intelligent intervention.

cs.RO↗

arXiv2026

Towards Sequence Modeling Alignment between Tokenizer and Autoregressive Model

Autoregressive image generation aims to predict the next token based on previous ones. However, this process is challenged by the bidirectional dependencies inherent in conventional image tokenizations, which creates a fundamental misalignment with the unidirectional nature of autoregressive models. To resolve this, we introduce AliTok, a novel Aligned Tokenizer that alters the dependency structure of the token sequence. AliTok employs a bidirectional encoder constrained by a causal decoder, a design that compels the encoder to produce a token sequence with both semantic richness and forward-dependency. Furthermore, by incorporating prefix tokens and employing a two-stage tokenizer training process to enhance reconstruction performance, AliTok achieves high fidelity and predictability simultaneously. Building upon AliTok, a standard decoder-only autoregressive model with just 177M parameters achieves a gFID of 1.44 and an IS of 319.5 on ImageNet-256. Scaling to 662M, our model reaches a gFID of 1.28, surpassing the SOTA diffusion method with 10x faster sampling. On ImageNet-512, our 318M model also achieves a SOTA gFID of 1.39. Code and weights at https://github.com/ali-vilab/alitok.

cs.CV↗

arXiv2025

SCALER: SAM-Enhanced Collaborative Learning for Label-Deficient Concealed Object Segmentation

Existing methods for label-deficient concealed object segmentation (LDCOS) either rely on consistency constraints or Segment Anything Model (SAM)-based pseudo-labeling. However, their performance remains limited due to the intrinsic concealment of targets and the scarcity of annotations. This study investigates two key questions: (1) Can consistency constraints and SAM-based supervision be jointly integrated to better exploit complementary information and enhance the segmenter? and (2) beyond that, can the segmenter in turn guide SAM through reciprocal supervision, enabling mutual improvement? To answer these questions, we present SCALER, a unified collaborative framework toward LDCOS that jointly optimizes a mean-teacher segmenter and a learnable SAM. SCALER operates in two alternating phases. In \textbf{Phase \uppercase\expandafter{\romannumeral1}}, the segmenter is optimized under fixed SAM supervision using entropy-based image-level and uncertainty-based pixel-level weighting to select reliable pseudo-label regions and emphasize harder examples. In \textbf{Phase \uppercase\expandafter{\romannumeral2}}, SAM is updated via augmentation invariance and noise resistance losses, leveraging its inherent robustness to perturbations. Experiments demonstrate that SCALER yields consistent performance gains across eight semi- and weakly-supervised COS tasks. The results further suggest that SCALER can serve as a general training paradigm to enhance both lightweight segmenters and large foundation models under label-scarce conditions. Code will be released.

cs.CV↗

arXiv2025

Exploiting Discriminative Codebook Prior for Autoregressive Image Generation

Advanced discrete token-based autoregressive image generation systems first tokenize images into sequences of token indices with a codebook, and then model these sequences in an autoregressive paradigm. While autoregressive generative models are trained only on index values, the prior encoded in the codebook, which contains rich token similarity information, is not exploited. Recent studies have attempted to incorporate this prior by performing naive k-means clustering on the tokens, helping to facilitate the training of generative models with a reduced codebook. However, we reveal that k-means clustering performs poorly in the codebook feature space due to inherent issues, including token space disparity and centroid distance inaccuracy. In this work, we propose the Discriminative Codebook Prior Extractor (DCPE) as an alternative to k-means clustering for more effectively mining and utilizing the token similarity information embedded in the codebook. DCPE replaces the commonly used centroid-based distance, which is found to be unsuitable and inaccurate for the token feature space, with a more reasonable instance-based distance. Using an agglomerative merging technique, it further addresses the token space disparity issue by avoiding splitting high-density regions and aggregating low-density ones. Extensive experiments demonstrate that DCPE is plug-and-play and integrates seamlessly with existing codebook prior-based paradigms. With the discriminative prior extracted, DCPE accelerates the training of autoregressive models by 42% on LlamaGen-B and improves final FID and IS performance.

cs.CV↗

arXiv2025

UniQA: Unified Vision-Language Pre-training for Image Quality and Aesthetic Assessment

Image Quality Assessment (IQA) and Image Aesthetic Assessment (IAA) aim to simulate human subjective perception of image visual quality and aesthetic appeal. Despite distinct learning objectives, they have underlying interconnectedness due to consistent human assessment perception. In this paper, we propose Unified vision-language pre-training of Quality and Aesthetics (UniQA}), to extract useful and common representations from two tasks, thereby benefiting them simultaneously. However, the lack of text in the IQA datasets and the textual noise in the IAA datasets pose severe challenges for multimodal pre-training. To address this, we (1) utilize multimodal large language models (MLLMs) to generate high-quality text descriptions; (2) use the generated text for IAA as metadata to purify noisy IAA data. To effectively adapt the pre-trained UniQA to downstream tasks, we further propose a lightweight adapter that utilizes versatile cues to fully exploit the extensive knowledge of the pre-trained model. UniQA demonstrates high competitiveness in various image assessment tasks, including classical IQA and IAA tasks, few-label IQA, and other downstream tasks, showing promise as a foundational assessment model. Codes are available at https://github.com/zht8506/UniQA.

cs.CV↗

arXiv2025

Gamma: Toward Generic Image Assessment with Mixture of Assessment Experts

Image assessment aims to evaluate the quality and aesthetics of images and has been applied across various scenarios, such as natural and AIGC scenes. Existing methods mostly address these sub-tasks or scenes individually. While some works attempt to develop unified image assessment models, they have struggled to achieve satisfactory performance or cover a broad spectrum of assessment scenarios. In this paper, we present \textbf{Gamma}, a \textbf{G}eneric im\textbf{A}ge assess\textbf{M}ent model using \textbf{M}ixture of \textbf{A}ssessment Experts, which can effectively assess images from diverse scenes through mixed-dataset training. Achieving unified training in image assessment presents significant challenges due to annotation biases across different datasets. To address this issue, we first propose a Mixture of Assessment Experts (MoAE) module, which employs shared and adaptive experts to dynamically learn common and specific knowledge for different datasets, respectively. In addition, we introduce a Scene-based Differential Prompt (SDP) strategy, which uses scene-specific prompts to provide prior knowledge and guidance during the learning process, further boosting adaptation for various scenes. Our Gamma model is trained and evaluated on 12 datasets spanning 6 image assessment scenarios. Extensive experiments show that our unified Gamma outperforms other state-of-the-art mixed-training methods by significant margins while covering more scenes. Codes are available at https://github.com/zht8506/Gamma.

cs.CV↗

arXiv2025

Segment Concealed Objects with Incomplete Supervision

Incompletely-Supervised Concealed Object Segmentation (ISCOS) involves segmenting objects that seamlessly blend into their surrounding environments, utilizing incompletely annotated data, such as weak and semi-annotations, for model training. This task remains highly challenging due to (1) the limited supervision provided by the incompletely annotated training data, and (2) the difficulty of distinguishing concealed objects from the background, which arises from the intrinsic similarities in concealed scenarios. In this paper, we introduce the first unified method for ISCOS to address these challenges. To tackle the issue of incomplete supervision, we propose a unified mean-teacher framework, SEE, that leverages the vision foundation model, ``\emph{Segment Anything Model (SAM)}'', to generate pseudo-labels using coarse masks produced by the teacher model as prompts. To mitigate the effect of low-quality segmentation masks, we introduce a series of strategies for pseudo-label generation, storage, and supervision. These strategies aim to produce informative pseudo-labels, store the best pseudo-labels generated, and select the most reliable components to guide the student model, thereby ensuring robust network training. Additionally, to tackle the issue of intrinsic similarity, we design a hybrid-granularity feature grouping module that groups features at different granularities and aggregates these results. By clustering similar features, this module promotes segmentation coherence, facilitating more complete segmentation for both single-object and multiple-object images. We validate the effectiveness of our approach across multiple ISCOS tasks, and experimental results demonstrate that our method achieves state-of-the-art performance. Furthermore, SEE can serve as a plug-and-play solution, enhancing the performance of existing models.

cs.CV↗

arXiv2025

Integrating Extra Modality Helps Segmentor Find Camouflaged Objects Well

Camouflaged Object Segmentation (COS) remains challenging because camouflaged objects exhibit only subtle visual differences from their backgrounds and single-modality RGB methods provide limited cues, leading researchers to explore multimodal data to improve segmentation accuracy. In this work, we presenet MultiCOS, a novel framework that effectively leverages diverse data modalities to improve segmentation performance. MultiCOS comprises two modules: Bi-space Fusion Segmentor (BFSer), which employs a state space and a latent space fusion mechanism to integrate cross-modal features within a shared representation and employs a fusion-feedback mechanism to refine context-specific features, and Cross-modal Knowledge Learner (CKLer), which leverages external multimodal datasets to generate pseudo-modal inputs and establish cross-modal semantic associations, transferring knowledge to COS models when real multimodal pairs are missing. When real multimodal COS data are unavailable, CKLer yields additional segmentation gains using only non-COS multimodal sources. Experiments on standard COS benchmarks show that BFSer outperforms existing multimodal baselines with both real and pseudo-modal data. Code will be released at \href{https://github.com/cnyvfang/MultiCOS}{GitHub}.

cs.CV↗

arXiv2025

UnfoldIR: Rethinking Deep Unfolding Network in Illumination Degradation Image Restoration

Deep unfolding networks (DUNs) are widely employed in illumination degradation image restoration (IDIR) to merge the interpretability of model-based approaches with the generalization of learning-based methods. However, the performance of DUN-based methods remains considerably inferior to that of state-of-the-art IDIR solvers. Our investigation indicates that this limitation does not stem from structural shortcomings of DUNs but rather from the limited exploration of the unfolding structure, particularly for (1) constructing task-specific restoration models, (2) integrating advanced network architectures, and (3) designing DUN-specific loss functions. To address these issues, we propose a novel DUN-based method, UnfoldIR, for IDIR tasks. UnfoldIR first introduces a new IDIR model with dedicated regularization terms for smoothing illumination and enhancing texture. We unfold the iterative optimized solution of this model into a multistage network, with each stage comprising a reflectance-assisted illumination correction (RAIC) module and an illumination-guided reflectance enhancement (IGRE) module. RAIC employs a visual state space (VSS) to extract non-local features, enforcing illumination smoothness, while IGRE introduces a frequency-aware VSS to globally align similar textures, enabling mildly degraded regions to guide the enhancement of details in more severely degraded areas. This suppresses noise while enhancing details. Furthermore, given the multistage structure, we propose an inter-stage information consistent loss to maintain network stability in the final stages. This loss contributes to structural preservation and sustains the model's performance even in unsupervised settings. Experiments verify our effectiveness across 5 IDIR tasks and 3 downstream problems.

cs.CV↗

arXiv2025

Real-world Image Dehazing with Coherence-based Pseudo Labeling and Cooperative Unfolding Network

Real-world Image Dehazing (RID) aims to alleviate haze-induced degradation in real-world settings. This task remains challenging due to the complexities in accurately modeling real haze distributions and the scarcity of paired real-world data. To address these challenges, we first introduce a cooperative unfolding network that jointly models atmospheric scattering and image scenes, effectively integrating physical knowledge into deep networks to restore haze-contaminated details. Additionally, we propose the first RID-oriented iterative mean-teacher framework, termed the Coherence-based Label Generator, to generate high-quality pseudo labels for network training. Specifically, we provide an optimal label pool to store the best pseudo-labels during network training, leveraging both global and local coherence to select high-quality candidates and assign weights to prioritize haze-free regions. We verify the effectiveness of our method, with experiments demonstrating that it achieves state-of-the-art performance on RID tasks. Code will be available at https://github.com/cnyvfang/CORUN-Colabator.

cs.CV↗

arXiv2025

UniAnimate-DiT: Human Image Animation with Large-Scale Video Diffusion Transformer

This report presents UniAnimate-DiT, an advanced project that leverages the cutting-edge and powerful capabilities of the open-source Wan2.1 model for consistent human image animation. Specifically, to preserve the robust generative capabilities of the original Wan2.1 model, we implement Low-Rank Adaptation (LoRA) technique to fine-tune a minimal set of parameters, significantly reducing training memory overhead. A lightweight pose encoder consisting of multiple stacked 3D convolutional layers is designed to encode motion information of driving poses. Furthermore, we adopt a simple concatenation operation to integrate the reference appearance into the model and incorporate the pose information of the reference image for enhanced pose alignment. Experimental results show that our approach achieves visually appearing and temporally consistent high-fidelity animations. Trained on 480p (832x480) videos, UniAnimate-DiT demonstrates strong generalization capabilities to seamlessly upscale to 720P (1280x720) during inference. The training and inference code is publicly available at https://github.com/ali-vilab/UniAnimate-DiT.

cs.CV↗

arXiv2025

Does Your Vision-Language Model Get Lost in the Long Video Sampling Dilemma?

The rise of Large Vision-Language Models (LVLMs) has significantly advanced video understanding. However, efficiently processing long videos remains a challenge due to the ``Sampling Dilemma'': low-density sampling risks missing critical information, while high-density sampling introduces redundancy. To address this issue, we introduce LSDBench, the first benchmark designed to evaluate LVLMs on long-video tasks by constructing high Necessary Sampling Density (NSD) questions, where NSD represents the minimum sampling density required to accurately answer a given question. LSDBench focuses on dense, short-duration actions to rigorously assess the sampling strategies employed by LVLMs. To tackle the challenges posed by high-NSD questions, we propose a novel Reasoning-Driven Hierarchical Sampling (RHS) framework, which combines global localization of question-relevant cues with local dense sampling for precise inference. Additionally, we develop a lightweight Semantic-Guided Frame Selector to prioritize informative frames, enabling RHS to achieve comparable or superior performance with significantly fewer sampled frames. Together, our LSDBench and RHS framework address the unique challenges of high-NSD long-video tasks, setting a new standard for evaluating and improving LVLMs in this domain. Our benchmark and evaluation codes has been released at: https://github.com/dvlab-research/LSDBench

cs.CV↗