Lefei Zhang

Jiaqi Ma, Tianheng Cheng, Guoli Wang et al.

Image restoration aims to reconstruct degraded images, e.g., denoising or deblurring. Existing works focus on designing task-specific methods and there are inadequate attempts at universal methods. However, simply unifying multiple tasks into one universal architecture suffers from uncontrollable and undesired predictions. To address those issues, we explore prompt learning in universal architectures for image restoration tasks. In this paper, we present Degradation-aware Visual Prompts, which encode various types of image degradation, e.g., noise and blur, into unified visual prompts. These degradation-aware prompts provide control over image processing and allow weighted combinations for customized image restoration. We then leverage degradation-aware visual prompts to establish a controllable and universal model for image restoration, called ProRes, which is applicable to an extensive range of image restoration tasks. ProRes leverages the vanilla Vision Transformer (ViT) without any task-specific designs. Furthermore, the pre-trained ProRes can easily adapt to new tasks through efficient prompt tuning with only a few images. Without bells and whistles, ProRes achieves competitive performance compared to task-specific methods and experiments can demonstrate its ability for controllable restoration and adaptation for new tasks. The code and models will be released in \url{https://github.com/leonmakise/ProRes}.

Yangyang Xu, Yibo Yang, Lefei Zhang

Convolution neural networks (CNNs) and Transformers have their own advantages and both have been widely used for dense prediction in multi-task learning (MTL). Most of the current studies on MTL solely rely on CNN or Transformer. In this work, we present a novel MTL model by combining both merits of deformable CNN and query-based Transformer for multi-task learning of dense prediction. Our method, named DeMT, is based on a simple and effective encoder-decoder architecture (i.e., deformable mixer encoder and task-aware transformer decoder). First, the deformable mixer encoder contains two types of operators: the channel-aware mixing operator leveraged to allow communication among different channels ($i.e.,$ efficient channel location mixing), and the spatial-aware deformable operator with deformable convolution applied to efficiently sample more informative spatial locations (i.e., deformed features). Second, the task-aware transformer decoder consists of the task interaction block and task query block. The former is applied to capture task interaction features via self-attention. The latter leverages the deformed features and task-interacted features to generate the corresponding task-specific feature through a query-based Transformer for corresponding task predictions. Extensive experiments on two dense image prediction datasets, NYUD-v2 and PASCAL-Context, demonstrate that our model uses fewer GFLOPs and significantly outperforms current Transformer- and CNN-based competitive models on a variety of metrics. The code are available at https://github.com/yangyangxu0/DeMT .

Yawei Li, Kai Zhang, Radu Timofte et al. · eth-zurich, tencent-ai

This paper reviews the NTIRE 2022 challenge on efficient single image super-resolution with focus on the proposed solutions and results. The task of the challenge was to super-resolve an input image with a magnification factor of $\times$4 based on pairs of low and corresponding high resolution images. The aim was to design a network for single image super-resolution that achieved improvement of efficiency measured according to several metrics including runtime, parameters, FLOPs, activations, and memory consumption while at least maintaining the PSNR of 29.00dB on DIV2K validation set. IMDN is set as the baseline for efficiency measurement. The challenge had 3 tracks including the main track (runtime), sub-track one (model complexity), and sub-track two (overall performance). In the main track, the practical runtime performance of the submissions was evaluated. The rank of the teams were determined directly by the absolute value of the average runtime on the validation set and test set. In sub-track one, the number of parameters and FLOPs were considered. And the individual rankings of the two metrics were summed up to determine a final ranking in this track. In sub-track two, all of the five metrics mentioned in the description of the challenge including runtime, parameter count, FLOPs, activations, and memory consumption were considered. Similar to sub-track one, the rankings of five metrics were summed up to determine a final ranking. The challenge had 303 registered participants, and 43 teams made valid submissions. They gauge the state-of-the-art in efficient single image super-resolution.

Yangyang Xu, Xiangtai Li, Haobo Yuan et al.

Previous multi-task dense prediction studies developed complex pipelines such as multi-modal distillations in multiple stages or searching for task relational contexts for each task. The core insight beyond these methods is to maximize the mutual effects of each task. Inspired by the recent query-based Transformers, we propose a simple pipeline named Multi-Query Transformer (MQTransformer) that is equipped with multiple queries from different tasks to facilitate the reasoning among multiple tasks and simplify the cross-task interaction pipeline. Instead of modeling the dense per-pixel context among different tasks, we seek a task-specific proxy to perform cross-task reasoning via multiple queries where each query encodes the task-related context. The MQTransformer is composed of three key components: shared encoder, cross-task query attention module and shared decoder. We first model each task with a task-relevant query. Then both the task-specific feature output by the feature extractor and the task-relevant query are fed into the shared encoder, thus encoding the task-relevant query from the task-specific feature. Secondly, we design a cross-task query attention module to reason the dependencies among multiple task-relevant queries; this enables the module to only focus on the query-level interaction. Finally, we use a shared decoder to gradually refine the image features with the reasoned query features from different tasks. Extensive experiment results on two dense prediction datasets (NYUD-v2 and PASCAL-Context) show that the proposed method is an effective approach and achieves state-of-the-art results. Code and models are available at https://github.com/yangyangxu0/MQTransformer.

Jiaqi Ma, Shengyuan Yan, Lefei Zhang et al.

In order to get raw images of high quality for downstream Image Signal Process (ISP), in this paper we present an Efficient Locally Multiplicative Transformer called ELMformer for raw image restoration. ELMformer contains two core designs especially for raw images whose primitive attribute is single-channel. The first design is a Bi-directional Fusion Projection (BFP) module, where we consider both the color characteristics of raw images and spatial structure of single-channel. The second one is that we propose a Locally Multiplicative Self-Attention (L-MSA) scheme to effectively deliver information from the local space to relevant parts. ELMformer can efficiently reduce the computational consumption and perform well on raw image restoration tasks. Enhanced by these two core designs, ELMformer achieves the highest performance and keeps the lowest FLOPs on raw denoising and raw deblurring benchmarks compared with state-of-the-arts. Extensive experiments demonstrate the superiority and generalization ability of ELMformer. On SIDD benchmark, our method has even better denoising performance than ISP-based methods which need huge amount of additional sRGB training images. The codes are release at https://github.com/leonmakise/ELMformer.

Shi Chen, Lefei Zhang, Liangpei Zhang

Hyperspectral image super-resolution has attained widespread prominence to enhance the spatial resolution of hyperspectral images. However, convolution-based methods have encountered challenges in harnessing the global spatial-spectral information. The prevailing transformer-based methods have not adequately captured the long-range dependencies in both spectral and spatial dimensions. To alleviate this issue, we propose a novel cross-scope spatial-spectral Transformer (CST) to efficiently investigate long-range spatial and spectral similarities for single hyperspectral image super-resolution. Specifically, we devise cross-attention mechanisms in spatial and spectral dimensions to comprehensively model the long-range spatial-spectral characteristics. By integrating global information into the rectangle-window self-attention, we first design a cross-scope spatial self-attention to facilitate long-range spatial interactions. Then, by leveraging appropriately characteristic spatial-spectral features, we construct a cross-scope spectral self-attention to effectively capture the intrinsic correlations among global spectral bands. Finally, we elaborate a concise feed-forward neural network to enhance the feature representation capacity in the Transformer structure. Extensive experiments over three hyperspectral datasets demonstrate that the proposed CST is superior to other state-of-the-art methods both quantitatively and visually. The code is available at \url{https://github.com/Tomchenshi/CST.git}.

Yuchun Miao, Lefei Zhang, Liangpei Zhang et al.

Diffusion models have recently received a surge of interest due to their impressive performance for image restoration, especially in terms of noise robustness. However, existing diffusion-based methods are trained on a large amount of training data and perform very well in-distribution, but can be quite susceptible to distribution shift. This is especially inappropriate for data-starved hyperspectral image (HSI) restoration. To tackle this problem, this work puts forth a self-supervised diffusion model for HSI restoration, namely Denoising Diffusion Spatio-Spectral Model (\texttt{DDS2M}), which works by inferring the parameters of the proposed Variational Spatio-Spectral Module (VS2M) during the reverse diffusion process, solely using the degraded HSI without any extra training data. In VS2M, a variational inference-based loss function is customized to enable the untrained spatial and spectral networks to learn the posterior distribution, which serves as the transitions of the sampling chain to help reverse the diffusion process. Benefiting from its self-supervised nature and the diffusion process, \texttt{DDS2M} enjoys stronger generalization ability to various HSIs compared to existing diffusion-based methods and superior robustness to noise compared to existing HSI restoration methods. Extensive experiments on HSI denoising, noisy HSI completion and super-resolution on a variety of HSIs demonstrate \texttt{DDS2M}'s superiority over the existing task-specific state-of-the-arts.

Yangyang Xu, Yibo Yang, Bernard Ghanem et al.

Convolution neural networks and Transformers have their own advantages and both have been widely used for dense prediction in multi-task learning (MTL). Existing studies typically employ either CNNs (effectively capture local spatial patterns) or Transformers (capturing long-range dependencies) independently, but integrating their strengths may yield more robust models. In this work, we present an efficient MTL model that combines the adaptive capabilities of deformable CNN and query-based Transformer with shared gating for MTL of dense prediction. This combination may offer a simple and efficient solution owing to its powerful and flexible task-specific learning and the advantages of lower cost, less complexity, and smaller parameters than traditional MTL methods. We introduce an efficient multi-task dense prediction with deformable and gating mixer (DeGMix). First, the deformable mixer encoder contains two types of operators: the channel-aware mixing operator leveraged to allow communication among different channels, and the spatial-aware deformable operator with deformable convolution applied to efficiently sample more informative spatial locations. Second, the task-aware gating transformer decoder is used to perform task-specific predictions, in which task interaction block integrated with self-attention is applied to capture task interaction features, and the task query block integrated with gating attention is leveraged to dynamically select the corresponding task-specific features. Furthermore, the results of the experiment demonstrate that the proposed DeGMix uses fewer GFLOPs and significantly outperforms current Transformer-based and CNN-based competitive models on a variety of metrics on three dense prediction datasets. Our code and models are available at https://github.com/yangyangxu0/DeMTG.

Cong Yang, Zuchao Li, Hongzan Jiao et al.

Recently, while significant progress has been made in remote sensing image change captioning, existing methods fail to filter out areas unrelated to actual changes, making models susceptible to irrelevant features. In this article, we propose a novel multimodal framework for remote sensing image change captioning, guided by Key Change Features and Instruction-tuned (KCFI). This framework aims to fully leverage the intrinsic knowledge of large language models through visual instructions and enhance the effectiveness and accuracy of change features using pixel-level change detection tasks. Specifically, KCFI includes a ViTs encoder for extracting bi-temporal remote sensing image features, a key feature perceiver for identifying critical change areas, a pixel-level change detection decoder to constrain key change features, and an instruction-tuned decoder based on a large language model. Moreover, to ensure that change description and change detection tasks are jointly optimized, we employ a dynamic weight-averaging strategy to balance the losses between the two tasks. We also explore various feature combinations for visual fine-tuning instructions and demonstrate that using only key change features to guide the large language model is the optimal choice. To validate the effectiveness of our approach, we compare it against several state-of-the-art change captioning methods on the LEVIR-CC dataset, achieving the best performance. Our code will be available at https://github.com/yangcong356/KCFI.git.

Zihong Zhang, Zuchao Li, Lefei Zhang et al.

Autoregressive decoding in Large Language Models (LLMs) generates one token per step, causing high inference latency. Speculative decoding (SD) mitigates this through a guess-and-verify strategy, but existing training-free variants face trade-offs: retrieval-based drafts break when no exact match exists, while logits-based drafts lack structural guidance. We propose $\textbf{RACER}$ ($\textbf{R}$etrieval-$\textbf{A}$ugmented $\textbf{C}$ont$\textbf{e}$xtual $\textbf{R}$apid Speculative Decoding), a lightweight and training-free method that integrates retrieved exact patterns with logit-driven future cues. This unification supplies both reliable anchors and flexible extrapolation, yielding richer speculative drafts. Experiments on Spec-Bench, HumanEval, and MGSM-ZH demonstrate that RACER consistently accelerates inference, achieving more than $2\times$ speedup over autoregressive decoding, and outperforms prior training-free methods, offering a scalable, plug-and-play solution for efficient LLM decoding. Our source code is available at $\href{https://github.com/hkr04/RACER}{https://github.com/hkr04/RACER}$.

Anke Tang, Li Shen, Yong Luo et al.

Deep model training on extensive datasets is increasingly becoming cost-prohibitive, prompting the widespread adoption of deep model fusion techniques to leverage knowledge from pre-existing models. From simple weight averaging to more sophisticated methods like AdaMerging, model fusion effectively improves model performance and accelerates the development of new models. However, potential interference between parameters of individual models and the lack of interpretability in the fusion progress remain significant challenges. Existing methods often try to resolve the parameter interference issue by evaluating attributes of parameters, such as their magnitude or sign, or by parameter pruning. In this study, we begin by examining the fine-tuning of linear layers through the lens of subspace analysis and explicitly define parameter interference as an optimization problem to shed light on this subject. Subsequently, we introduce an innovative approach to model fusion called zero-shot Sparse MIxture of Low-rank Experts (SMILE) construction, which allows for the upscaling of source models into an MoE model without extra data or further training. Our approach relies on the observation that fine-tuning mostly keeps the important parts from the pre-training, but it uses less significant or unused areas to adapt to new tasks. Also, the issue of parameter interference, which is intrinsically intractable in the original parameter space, can be managed by expanding the dimensions. We conduct extensive experiments across diverse scenarios, such as image classification and text generation tasks, using full fine-tuning and LoRA fine-tuning, and we apply our method to large language models (CLIP models, Flan-T5 models, and Mistral-7B models), highlighting the adaptability and scalability of SMILE. Code is available at https://github.com/tanganke/fusion_bench

Yibo Yang, Haobo Yuan, Xiangtai Li et al.

How to enable learnability for new classes while keeping the capability well on old classes has been a crucial challenge for class incremental learning. Beyond the normal case, long-tail class incremental learning and few-shot class incremental learning are also proposed to consider the data imbalance and data scarcity, respectively, which are common in real-world implementations and further exacerbate the well-known problem of catastrophic forgetting. Existing methods are specifically proposed for one of the three tasks. In this paper, we offer a unified solution to the misalignment dilemma in the three tasks. Concretely, we propose neural collapse terminus that is a fixed structure with the maximal equiangular inter-class separation for the whole label space. It serves as a consistent target throughout the incremental training to avoid dividing the feature space incrementally. For CIL and LTCIL, we further propose a prototype evolving scheme to drive the backbone features into our neural collapse terminus smoothly. Our method also works for FSCIL with only minor adaptations. Theoretical analysis indicates that our method holds the neural collapse optimality in an incremental fashion regardless of data imbalance or data scarcity. We also design a generalized case where we do not know the total number of classes and whether the data distribution is normal, long-tail, or few-shot for each coming session, to test the generalizability of our method. Extensive experiments with multiple datasets are conducted to demonstrate the effectiveness of our unified solution to all the three tasks and the generalized case.

Tianshuo Peng, Zuchao Li, Lefei Zhang et al.

Universal Information Extraction (UIE) has been introduced as a unified framework for various Information Extraction (IE) tasks and has achieved widespread success. Despite this, UIE models have limitations. For example, they rely heavily on span boundaries in the data during training, which does not reflect the reality of span annotation challenges. Slight adjustments to positions can also meet requirements. Additionally, UIE models lack attention to the limited span length feature in IE. To address these deficiencies, we propose the Fuzzy Span Universal Information Extraction (FSUIE) framework. Specifically, our contribution consists of two concepts: fuzzy span loss and fuzzy span attention. Our experimental results on a series of main IE tasks show significant improvement compared to the baseline, especially in terms of fast convergence and strong performance with small amounts of data and training epochs. These results demonstrate the effectiveness and generalization of FSUIE in different tasks, settings, and scenarios.

Yineng Chen, Zuchao Li, Lefei Zhang et al.

Optimizer is an essential component for the success of deep learning, which guides the neural network to update the parameters according to the loss on the training set. SGD and Adam are two classical and effective optimizers on which researchers have proposed many variants, such as SGDM and RAdam. In this paper, we innovatively combine the backward-looking and forward-looking aspects of the optimizer algorithm and propose a novel \textsc{Admeta} (\textbf{A} \textbf{D}ouble exponential \textbf{M}oving averag\textbf{E} \textbf{T}o \textbf{A}daptive and non-adaptive momentum) optimizer framework. For backward-looking part, we propose a DEMA variant scheme, which is motivated by a metric in the stock market, to replace the common exponential moving average scheme. While in the forward-looking part, we present a dynamic lookahead strategy which asymptotically approaches a set value, maintaining its speed at early stage and high convergence performance at final stage. Based on this idea, we provide two optimizer implementations, \textsc{AdmetaR} and \textsc{AdmetaS}, the former based on RAdam and the latter based on SGDM. Through extensive experiments on diverse tasks, we find that the proposed \textsc{Admeta} optimizer outperforms our base optimizers and shows advantages over recently proposed competitive optimizers. We also provide theoretical proof of these two algorithms, which verifies the convergence of our proposed \textsc{Admeta}.

Meng Lan, Jing Zhang, Lefei Zhang et al.

Recently, the joint learning framework (JOINT) integrates matching based transductive reasoning and online inductive learning to achieve accurate and robust semi-supervised video object segmentation (SVOS). However, using the mask embedding as the label to guide the generation of target features in the two branches may result in inadequate target representation and degrade the performance. Besides, how to reasonably fuse the target features in the two different branches rather than simply adding them together to avoid the adverse effect of one dominant branch has not been investigated. In this paper, we propose a novel framework that emphasizes Learning to Learn Better (LLB) target features for SVOS, termed LLB, where we design the discriminative label generation module (DLGM) and the adaptive fusion module to address these issues. Technically, the DLGM takes the background-filtered frame instead of the target mask as input and adopts a lightweight encoder to generate the target features, which serves as the label of the online few-shot learner and the value of the decoder in the transformer to guide the two branches to learn more discriminative target representation. The adaptive fusion module maintains a learnable gate for each branch, which reweighs the element-wise feature representation and allows an adaptive amount of target information in each branch flowing to the fused target feature, thus preventing one branch from being dominant and making the target feature more robust to distractor. Extensive experiments on public benchmarks show that our proposed LLB method achieves state-of-the-art performance.

Meng Lan, Fu Rong, Zuchao Li et al.

Referring video object segmentation (RVOS) aims to segment the target object in a video sequence described by a language expression. Typical multimodal Transformer based RVOS approaches process video sequence in a frame-independent manner to reduce the high computational cost, which however restricts the performance due to the lack of inter-frame interaction for temporal coherence modeling and spatio-temporal representation learning of the referred object. Besides, the absence of sufficient cross-modal interactions results in weak correlation between the visual and linguistic features, which increases the difficulty of decoding the target information and limits the performance of the model. In this paper, we propose a bidirectional correlation-driven inter-frame interaction Transformer, dubbed BIFIT, to address these issues in RVOS. Specifically, we design a lightweight and plug-and-play inter-frame interaction module in the Transformer decoder to efficiently learn the spatio-temporal features of the referred object, so as to decode the object information in the video sequence more precisely and generate more accurate segmentation results. Moreover, a bidirectional vision-language interaction module is implemented before the multimodal Transformer to enhance the correlation between the visual and linguistic features, thus facilitating the language queries to decode more precise object information from visual features and ultimately improving the segmentation performance. Extensive experimental results on four benchmarks validate the superiority of our BIFIT over state-of-the-art methods and the effectiveness of our proposed modules.

Luohe Shi, Yao Yao, Zuchao Li et al.

Large language models (LLMs) have rapidly advanced and demonstrated impressive capabilities. In-Context Learning (ICL) and Parameter-Efficient Fine-Tuning (PEFT) are currently two mainstream methods for augmenting LLMs to downstream tasks. ICL typically constructs a few-shot learning scenario, either manually or by setting up a Retrieval-Augmented Generation (RAG) system, helping models quickly grasp domain knowledge or question-answering patterns without changing model parameters. However, this approach involves trade-offs, such as slower inference speed and increased space occupancy. PEFT assists the model in adapting to tasks through minimal parameter modifications, but the training process still demands high hardware requirements, even with a small number of parameters involved. To address these challenges, we propose Reference Trustable Decoding (RTD), a paradigm that allows models to quickly adapt to new tasks without fine-tuning, maintaining low inference costs. RTD constructs a reference datastore from the provided training examples and optimizes the LLM's final vocabulary distribution by flexibly selecting suitable references based on the input, resulting in more trustable responses and enabling the model to adapt to downstream tasks at a low cost. Experimental evaluations on various LLMs using different benchmarks demonstrate that RTD establishes a new paradigm for augmenting models to downstream tasks. Furthermore, our method exhibits strong orthogonality with traditional methods, allowing for concurrent usage. Our code can be found at https://github.com/ShiLuohe/ReferenceTrustableDecoding

Jialv Zou, Shaoyu Chen, Bencheng Liao et al.

Generative diffusion models for end-to-end autonomous driving often suffer from mode collapse, tending to generate conservative and homogeneous behaviors. While DiffusionDrive employs predefined anchors representing different driving intentions to partition the action space and generate diverse trajectories, its reliance on imitation learning lacks sufficient constraints, resulting in a dilemma between diversity and consistent high quality. In this work, we propose DiffusionDriveV2, which leverages reinforcement learning to both constrain low-quality modes and explore for superior trajectories. This significantly enhances the overall output quality while preserving the inherent multimodality of its core Gaussian Mixture Model. First, we use scale-adaptive multiplicative noise, ideal for trajectory planning, to promote broad exploration. Second, we employ intra-anchor GRPO to manage advantage estimation among samples generated from a single anchor, and inter-anchor truncated GRPO to incorporate a global perspective across different anchors, preventing improper advantage comparisons between distinct intentions (e.g., turning vs. going straight), which can lead to further mode collapse. DiffusionDriveV2 achieves 91.2 PDMS on the NAVSIM v1 dataset and 85.5 EPDMS on the NAVSIM v2 dataset in closed-loop evaluation with an aligned ResNet-34 backbone, setting a new record. Further experiments validate that our approach resolves the dilemma between diversity and consistent high quality for truncated diffusion models, achieving the best trade-off. Code and model will be available at https://github.com/hustvl/DiffusionDriveV2

Ruiheng Zhang, Jingfeng Yao, Huangxuan Zhao et al.

Despite recent progress, medical foundation models still struggle to unify visual understanding and generation, as these tasks have inherently conflicting goals: semantic abstraction versus pixel-level reconstruction. Existing approaches, typically based on parameter-shared autoregressive architectures, frequently lead to compromised performance in one or both tasks. To address this, we present UniX, a next-generation unified medical foundation model for chest X-ray understanding and generation. UniX decouples the two tasks into an autoregressive branch for understanding and a diffusion branch for high-fidelity generation. Crucially, a cross-modal self-attention mechanism is introduced to dynamically guide the generation process with understanding features. Coupled with a rigorous data cleaning pipeline and a multi-stage training strategy, this architecture enables synergistic collaboration between tasks while leveraging the strengths of diffusion models for superior generation. On two representative benchmarks, UniX achieves a 46.1% improvement in understanding performance (Micro-F1) and a 24.2% gain in generation quality (FD-RadDino), using only a quarter of the parameters of LLM-CXR. By achieving performance on par with task-specific models, our work establishes a scalable paradigm for synergistic medical image understanding and generation. Codes and models are available at https://github.com/ZrH42/UniX.

Xin Yan, Zuchao Li, Lefei Zhang

Masked Image Modeling (MIM) is a new self-supervised vision pre-training paradigm using a Vision Transformer (ViT). Previous works can be pixel-based or token-based, using original pixels or discrete visual tokens from parametric tokenizer models, respectively. Our proposed centroid-based approach, CCViT, leverages k-means clustering to obtain centroids for image modeling without supervised training of the tokenizer model, which only takes seconds to create. This non-parametric centroid tokenizer only takes seconds to create and is faster for token inference. The centroids can represent both patch pixels and index tokens with the property of local invariance. Specifically, we adopt patch masking and centroid replacing strategies to construct corrupted inputs, and two stacked encoder blocks to predict corrupted patch tokens and reconstruct original patch pixels. Experiments show that our CCViT achieves 84.4% top-1 accuracy on ImageNet-1K classification with ViT-B and 86.0% with ViT-L. We also transfer our pre-trained model to other downstream tasks. Our approach achieves competitive results with recent baselines without external supervision and distillation training from other models.

Ziyi Zhang, Li Shen, Deheng Ye et al.

Text-to-multiview (T2MV) diffusion models have shown great promise in generating multiple views of a scene from a single text prompt. While few-step backbones enable real-time T2MV generation, they often compromise key aspects of generation quality, such as per-view fidelity and cross-view consistency. Reinforcement learning (RL) finetuning offers a potential solution, yet existing approaches designed for single-image diffusion do not readily extend to the few-step T2MV setting, as they neglect cross-view coordination and suffer from weak learning signals in few-step regimes. To address this, we propose MVC-ZigAL, a tailored RL finetuning framework for few-step T2MV diffusion models. Specifically, its core insights are: (1) a new MDP formulation that jointly models all generated views and assesses their collective quality via a joint-view reward; (2) a novel advantage learning strategy that exploits the performance gains of a self-refinement sampling scheme over standard sampling, yielding stronger learning signals for effective RL finetuning; and (3) a unified RL framework that extends advantage learning with a Lagrangian dual formulation for multiview-constrained optimization, balancing single-view and joint-view objectives through adaptive primal-dual updates under a self-paced threshold curriculum that harmonizes exploration and constraint enforcement. Collectively, these designs enable robust and balanced RL finetuning for few-step T2MV diffusion models, yielding substantial gains in both per-view fidelity and cross-view consistency. Code is available at https://github.com/ZiyiZhang27/MVC-ZigAL.

Jiani Guo, Zuchao Li, Jie Wu et al.

Large Language Models (LLMs), constrained by limited context windows, often face significant performance degradation when reasoning over long contexts. To address this, Retrieval-Augmented Generation (RAG) retrieves and reasons over chunks but frequently sacrifices logical coherence due to its reliance on similarity-based rankings. Similarly, divide-and-conquer frameworks (DCF) split documents into small chunks for independent reasoning and aggregation. While effective for local reasoning, DCF struggles to capture long-range dependencies and risks inducing conflicts by processing chunks in isolation. To overcome these limitations, we propose ToM, a novel Tree-oriented MapReduce framework for long-context reasoning. ToM leverages the inherent hierarchical structure of long documents (e.g., main headings and subheadings) by constructing a DocTree through hierarchical semantic parsing and performing bottom-up aggregation. Using a Tree MapReduce approach, ToM enables recursive reasoning: in the Map step, rationales are generated at child nodes; in the Reduce step, these rationales are aggregated across sibling nodes to resolve conflicts or reach consensus at parent nodes. Experimental results on 70B+ LLMs show that ToM significantly outperforms existing divide-and-conquer frameworks and retrieval-augmented generation methods, achieving better logical coherence and long-context reasoning. Our code is available at https://github.com/gjn12-31/ToM .

Haoshen Wang, Lei Chen, Wei-Hua Zhang et al.

Cone beam computed tomography (CBCT)-guided puncture has become an established approach for diagnosing and treating early- to mid-stage thoracic tumours, yet the associated radiation exposure substantially elevates the risk of secondary malignancies. Although multiple low-dose CBCT strategies have been introduced, none have undergone validation using large-scale multicenter retrospective datasets, and prospective clinical evaluation remains lacking. Here, we propose DeepPriorCBCT - a three-stage deep learning framework that achieves diagnostic-grade reconstruction using only one-sixth of the conventional radiation dose. 4102 patients with 8675 CBCT scans from 12 centers were included to develop and validate DeepPriorCBCT. Additionally, a prospective cross-over trial (Registry number: NCT07035977) which recruited 138 patients scheduled for percutaneous thoracic puncture was conducted to assess the model's clinical applicability. Assessment by 11 physicians confirmed that reconstructed images were indistinguishable from original scans. Moreover, diagnostic performance and overall image quality were comparable to those generated by standard reconstruction algorithms. In the prospective trial, five radiologists reported no significant differences in image quality or lesion assessment between DeepPriorCBCT and the clinical standard (all P>0.05). Likewise, 25 interventionalists expressed no preference between model-based and full-sampling images for surgical guidance (Kappa<0.2). Radiation exposure with DeepPriorCBCT was reduced to approximately one-sixth of that with the conventional approach, and collectively, the findings confirm that it enables high-quality CBCT reconstruction under sparse sampling conditions while markedly decreasing intraoperative radiation risk.

Gaoren Lin, Huangxuan Zhao, Yuan Xiong et al.

Text-guided medical segmentation enhances segmentation accuracy by utilizing clinical reports as auxiliary information. However, existing methods typically rely on unaligned image and text encoders, which necessitate complex interaction modules for multimodal fusion. While CLIP provides a pre-aligned multimodal feature space, its direct application to medical imaging is limited by three main issues: insufficient preservation of fine-grained anatomical structures, inadequate modeling of complex clinical descriptions, and domain-specific semantic misalignment. To tackle these challenges, we propose TGC-Net, a CLIP-based framework focusing on parameter-efficient, task-specific adaptations. Specifically, it incorporates a Semantic-Structural Synergy Encoder (SSE) that augments CLIP's ViT with a CNN branch for multi-scale structural refinement, a Domain-Augmented Text Encoder (DATE) that injects large-language-model-derived medical knowledge, and a Vision-Language Calibration Module (VLCM) that refines cross-modal correspondence in a unified feature space. Experiments on five datasets across chest X-ray and thoracic CT modalities demonstrate that TGC-Net achieves state-of-the-art performance with substantially fewer trainable parameters, including notable Dice gains on challenging benchmarks.

Xu Zhang, Jiaqi Ma, Guoli Wang et al.

Existing All-in-One image restoration methods often fail to perceive degradation types and severity levels simultaneously, overlooking the importance of fine-grained quality perception. Moreover, these methods often utilize highly customized backbones, which hinder their adaptability and integration into more advanced restoration networks. To address these limitations, we propose Perceive-IR, a novel backbone-agnostic All-in-One image restoration framework designed for fine-grained quality control across various degradation types and severity levels. Its modular structure allows core components to function independently of specific backbones, enabling seamless integration into advanced restoration models without significant modifications. Specifically, Perceive-IR operates in two key stages: 1) multi-level quality-driven prompt learning stage, where a fine-grained quality perceiver is meticulously trained to discern three tier quality levels by optimizing the alignment between prompts and images within the CLIP perception space. This stage ensures a nuanced understanding of image quality, laying the groundwork for subsequent restoration; 2) restoration stage, where the quality perceiver is seamlessly integrated with a difficulty-adaptive perceptual loss, forming a quality-aware learning strategy. This strategy not only dynamically differentiates sample learning difficulty but also achieves fine-grained quality control by driving the restored image toward the ground truth while pulling it away from both low- and medium-quality samples.

Yapeng Li, Yong Luo, Lefei Zhang et al.

Transformer has been extensively explored for hyperspectral image (HSI) classification. However, transformer poses challenges in terms of speed and memory usage because of its quadratic computational complexity. Recently, the Mamba model has emerged as a promising approach, which has strong long-distance modeling capabilities while maintaining a linear computational complexity. However, representing the HSI is challenging for the Mamba due to the requirement for an integrated spatial and spectral understanding. To remedy these drawbacks, we propose a novel HSI classification model based on a Mamba model, named MambaHSI, which can simultaneously model long-range interaction of the whole image and integrate spatial and spectral information in an adaptive manner. Specifically, we design a spatial Mamba block (SpaMB) to model the long-range interaction of the whole image at the pixel-level. Then, we propose a spectral Mamba block (SpeMB) to split the spectral vector into multiple groups, mine the relations across different spectral groups, and extract spectral features. Finally, we propose a spatial-spectral fusion module (SSFM) to adaptively integrate spatial and spectral features of a HSI. To our best knowledge, this is the first image-level HSI classification model based on the Mamba. We conduct extensive experiments on four diverse HSI datasets. The results demonstrate the effectiveness and superiority of the proposed model for HSI classification. This reveals the great potential of Mamba to be the next-generation backbone for HSI models. Codes are available at https://github.com/li-yapeng/MambaHSI .

Anke Tang, Li Shen, Yong Luo et al.

Merging various task-specific Transformer-based models trained on different tasks into a single unified model can execute all the tasks concurrently. Previous methods, exemplified by task arithmetic, have been proven to be both effective and scalable. Existing methods have primarily focused on seeking a static optimal solution within the original model parameter space. A notable challenge is mitigating the interference between parameters of different models, which can substantially deteriorate performance. In this paper, we propose to merge most of the parameters while upscaling the MLP of the Transformer layers to a weight-ensembling mixture of experts (MoE) module, which can dynamically integrate shared and task-specific knowledge based on the input, thereby providing a more flexible solution that can adapt to the specific needs of each instance. Our key insight is that by identifying and separating shared knowledge and task-specific knowledge, and then dynamically integrating them, we can mitigate the parameter interference problem to a great extent. We conduct the conventional multi-task model merging experiments and evaluate the generalization and robustness of our method. The results demonstrate the effectiveness of our method and provide a comprehensive understanding of our method. The code is available at https://github.com/tanganke/weight-ensembling_MoE

Xiangyu Ma, Teng Xiao, Zuchao Li et al.

Diffusion models promise efficient parallel text generation but rely on bidirectional attention, creating a structural mismatch with pre-trained Autoregressive (AR) models. This incompatibility precludes reusing robust AR priors, necessitating prohibitive pre-training from scratch. To bridge this gap, we propose FLUID, a framework that efficiently adapts AR backbones to the diffusion paradigm. By enforcing Strictly Causal Alignment, FLUID enables seamless initialization from standard GPT-style checkpoints, circumventing the need for massive pre-training. Furthermore, we introduce Elastic Horizons, an entropy-driven mechanism that dynamically modulates denoising strides based on local information density rather than fixed schedules. Experiments demonstrate that FLUID achieves state-of-the-art performance while reducing training costs by orders of magnitude, effectively reconciling established AR foundations with efficient parallel generation. Our code is available at https://github.com/Oli-lab-nun/FLUID/tree/main.

Yuchun Miao, Sen Zhang, Yuqi Zhang et al.

Reinforcement Learning with Verifiable Rewards (RLVR) has become the dominant paradigm for advanced reasoning in Large Language Models (LLMs), but rollout samples are expensive to obtain, making sample efficiency a critical bottleneck. A natural remedy is to reuse each rollout batch for multiple gradient updates, a standard practice in classical RL. Yet in RLVR, this amplifies policy shift, leading to severe performance degradation. Detecting the onset of degradation early enough to stop reuse remains an open and challenging problem. We close this gap by identifying the \textit{Disproportionate Weight Divergence (DWD)} phenomenon: performance degradation is synchronized with a sharp surge in the \texttt{lm\_head} weight change, while intermediate layers remain stable. Empirically, we verify that DWD emerges consistently across diverse LLMs and tasks. Theoretically, we prove that (i) harmful gradients concentrate at the \texttt{lm\_head} while intermediate layers are structurally attenuated, and (ii) the \texttt{lm\_head} gradient norm lower-bounds the policy divergence. These results establish the \texttt{lm\_head} gradient norm as a principled, real-time signal of catastrophic policy shift. Guided by this insight, we propose \textit{Dynamic Gradient Gating (DGG)}, a lightweight intervention that monitors the \texttt{lm\_head} gradient norm in real time and intercepts harmful gradients before they corrupt the optimizer. DGG consistently matches or exceeds the standard single-use baseline, achieving up to $2.93\times$ sample efficiency and $2.14\times$ wall-clock speedup across math, ALFWorld, WebShop, and search-augmented QA tasks.

Xu Zhang, Huan Zhang, Guoli Wang et al.

All-in-One Image Restoration (AiOIR) has advanced significantly, offering promising solutions for complex real-world degradations. However, most existing approaches rely heavily on degradation-specific representations, often resulting in oversmoothing and artifacts. To address this, we propose ClearAIR, a novel AiOIR framework inspired by Human Visual Perception (HVP) and designed with a hierarchical, coarse-to-fine restoration strategy. First, leveraging the global priority of early HVP, we employ a Multimodal Large Language Model (MLLM)-based Image Quality Assessment (IQA) model for overall evaluation. Unlike conventional IQA, our method integrates cross-modal understanding to more accurately characterize complex, composite degradations. Building upon this overall assessment, we then introduce a region awareness and task recognition pipeline. A semantic cross-attention, leveraging semantic guidance unit, first produces coarse semantic prompts. Guided by this regional context, a degradation-aware module implicitly captures region-specific degradation characteristics, enabling more precise local restoration. Finally, to recover fine details, we propose an internal clue reuse mechanism. It operates in a self-supervised manner to mine and leverage the intrinsic information of the image itself, substantially enhancing detail restoration. Experimental results show that ClearAIR achieves superior performance across diverse synthetic and real-world datasets.

Weixi Song, Zuchao Li, Lefei Zhang et al.

With the prevalence of pre-training-fine-tuning paradigm, how to efficiently adapt the pre-trained model to the downstream tasks has been an intriguing issue. Parameter-Efficient Fine-Tuning (PEFT) methods have been proposed for low-cost adaptation. Although PEFT has demonstrated effectiveness and been widely applied, the underlying principles are still unclear. In this paper, we adopt the PAC-Bayesian generalization error bound, viewing pre-training as a shift of prior distribution which leads to a tighter bound for generalization error. We validate this shift from the perspectives of oscillations in the loss landscape and the quasi-sparsity in gradient distribution. Based on this, we propose a gradient-based sparse fine-tuning algorithm, named Sparse Increment Fine-Tuning (SIFT), and validate its effectiveness on a range of tasks including the GLUE Benchmark and Instruction-tuning. The code is accessible at https://github.com/song-wx/SIFT/.

Shuhan Xu, Siyuan Liang, Hongling Zheng et al.

Diffusion-based image-to-video (I2V) models increasingly exhibit world-model-like properties by implicitly capturing temporal dynamics. However, existing studies have mainly focused on visual quality and controllability, and the robustness of the state transition learned by the model remains understudied. To fill this gap, we are the first to analyze the vulnerability of I2V models, find that temporal control mechanisms constitute a new attack surface, and reveal the challenge of modeling them uniformly under different attack settings. Based on this, we propose a trajectory-control attack, called CtrlAttack, to interfere with state evolution during the generation process. Specifically, we represent the perturbation as a low-dimensional velocity field and construct a continuous displacement field via temporal integration, thereby affecting the model's state transitions while maintaining temporal consistency; meanwhile, we map the perturbation to the observation space, making the method applicable to both white-box and black-box attack settings. Experimental results show that even under low-dimensional and strongly regularized perturbation constraints, our method can still significantly disrupt temporal consistency by increasing the attack success rate (ASR) to over 90% in the white-box setting and over 80% in the black-box setting, while keeping the variation of the FID and FVD within 6 and 130, respectively, thus revealing the potential security risk of I2V models at the level of state dynamics.

Jiaqi Chen, Xiaoye Zhu, Tianyang Liu et al.

Large Language Models (LLMs) have revolutionized text generation, making detecting machine-generated text increasingly challenging. Although past methods have achieved good performance on detecting pure machine-generated text, those detectors have poor performance on distinguishing machine-revised text (rewriting, expansion, and polishing), which can have only minor changes from its original human prompt. As the content of text may originate from human prompts, detecting machine-revised text often involves identifying distinctive machine styles, e.g., worded favored by LLMs. However, existing methods struggle to detect machine-style phrasing hidden within the content contributed by humans. We propose the "Imitate Before Detect" (ImBD) approach, which first imitates the machine-style token distribution, and then compares the distribution of the text to be tested with the machine-style distribution to determine whether the text has been machine-revised. To this end, we introduce style preference optimization (SPO), which aligns a scoring LLM model to the preference of text styles generated by machines. The aligned scoring model is then used to calculate the style-conditional probability curvature (Style-CPC), quantifying the log probability difference between the original and conditionally sampled texts for effective detection. We conduct extensive comparisons across various scenarios, encompassing text revisions by six LLMs, four distinct text domains, and three machine revision types. Compared to existing state-of-the-art methods, our method yields a 13% increase in AUC for detecting text revised by open-source LLMs, and improves performance by 5% and 19% for detecting GPT-3.5 and GPT-4o revised text, respectively. Notably, our method surpasses the commercially trained GPT-Zero with just $1,000$ samples and five minutes of SPO, demonstrating its efficiency and effectiveness.

Luohe Shi, Zuchao Li, Lefei Zhang et al.

Large language models (LLMs) based on Transformer Decoders have become the preferred choice for conversational generative AI. Despite the overall superiority of the Decoder architecture, the gradually increasing Key-Value (KV) cache during inference has emerged as a primary efficiency bottleneck, both in aspects of memory consumption and data transfer bandwidth limitations. To address these challenges, we propose a paradigm called KV-Latent. By down-sampling the Key-Value vector dimensions into a latent space, we can significantly reduce the KV Cache footprint and improve inference speed, only with a small amount of extra training, less than 1\% of pre-training takes. Besides, we enhanced the stability of Rotary Positional Embedding applied on lower-dimensional vectors by modifying its frequency sampling mechanism, avoiding noise introduced by higher frequencies while retaining position attenuation. Our experiments, including both models with Grouped Query Attention and those without, have yielded satisfactory results. Finally, we conducted comparative experiments to study the impact of separately reducing Key and Value components on model's performance. Our approach allows for the construction of more efficient language model systems, and opens the new possibility on KV Cache saving and efficient LLMs. Our code is available at https://github.com/ShiLuohe/KV-Latent.

Quanwei Liu, Yanni Dong, Tao Huang et al.

Hyperspectral image (HSI) classification techniques have been intensively studied and a variety of models have been developed. However, these HSI classification models are confined to pocket models and unrealistic ways of dataset partitioning. The former limits the generalization performance of the model and the latter is partitioned leading to inflated model evaluation metrics, which results in plummeting model performance in the real world. Therefore, we propose a universal knowledge embedded contrastive learning framework (KnowCL) for supervised, unsupervised, and semisupervised HSI classification, which largely closes the gap between HSI classification models between pocket models and standard vision backbones. We present a new HSI processing pipeline in conjunction with a range of data transformation and augmentation techniques that provide diverse data representations and realistic data partitioning. The proposed framework based on this pipeline is compatible with all kinds of backbones and can fully exploit labeled and unlabeled samples with the expected training time. Furthermore, we design a new loss function, which can adaptively fuse the supervised loss and unsupervised loss, enhancing the learning performance. This proposed new classification paradigm shows great potential in exploring for HSI classification technology. The code can be accessed at \url{https://github.com/quanweiliu/KnowCL}.

Fu Rong, Meng Lan, Qian Zhang et al.

Referring video object segmentation (RVOS) aims to segment objects in a video according to textual descriptions, which requires the integration of multimodal information and temporal dynamics perception. The Segment Anything Model 2 (SAM 2) has shown great effectiveness across various video segmentation tasks. However, its application to offline RVOS is challenged by the translation of the text into effective prompts and a lack of global context awareness. In this paper, we propose a novel RVOS framework, termed MPG-SAM 2, to address these challenges. Specifically, MPG-SAM 2 employs a unified multimodal encoder to jointly encode video and textual features, generating semantically aligned video and text embeddings, along with multimodal class tokens. A mask prior generator utilizes the video embeddings and class tokens to create pseudo masks of target objects and global context. These masks are fed into the prompt encoder as dense prompts along with multimodal class tokens as sparse prompts to generate accurate prompts for SAM 2. To provide the online SAM 2 with a global view, we introduce a hierarchical global-historical aggregator, which allows SAM 2 to aggregate global and historical information of target objects at both pixel and object levels, enhancing the target representation and temporal consistency. Extensive experiments on several RVOS benchmarks demonstrate the superiority of MPG-SAM 2 and the effectiveness of our proposed modules. The code is available at https://github.com/rongfu-dsb/MPG-SAM2.

Xiangke Zeng, Zuchao Li, Lefei Zhang et al.

Chinese Spelling Correction (CSC) stands as a foundational Natural Language Processing (NLP) task, which primarily focuses on the correction of erroneous characters in Chinese texts. Certain existing methodologies opt to disentangle the error correction process, employing an additional error detector to pinpoint error positions. However, owing to the inherent performance limitations of error detector, precision and recall are like two sides of the coin which can not be both facing up simultaneously. Furthermore, it is also worth investigating how the error position information can be judiciously applied to assist the error correction. In this paper, we introduce a novel approach based on error detector-corrector framework. Our detector is designed to yield two error detection results, each characterized by high precision and recall. Given that the occurrence of errors is context-dependent and detection outcomes may be less precise, we incorporate the error detection results into the CSC task using an innovative feature fusion strategy and a selective masking strategy. Empirical experiments conducted on mainstream CSC datasets substantiate the efficacy of our proposed method.

Shengkai Hu, Jiaqi Ma, Jun Wan et al.

All-in-One Image Restoration (AiOIR) aims to recover high-quality images from diverse degradations within a unified framework. However, existing methods often fail to explicitly model degradation types and struggle to adapt their restoration behavior to complex or mixed degradations. To address these issues, we propose ClusIR, a Cluster-Guided Image Restoration framework that explicitly models degradation semantics through learnable clustering and propagates cluster-aware cues across spatial and frequency domains for adaptive restoration. Specifically, ClusIR comprises two key components: a Probabilistic Cluster-Guided Routing Mechanism (PCGRM) and a Degradation-Aware Frequency Modulation Module (DAFMM). The proposed PCGRM disentangles degradation recognition from expert activation, enabling discriminative degradation perception and stable expert routing. Meanwhile, DAFMM leverages the cluster-guided priors to perform adaptive frequency decomposition and targeted modulation, collaboratively refining structural and textural representations for higher restoration fidelity. The cluster-guided synergy seamlessly bridges semantic cues with frequency-domain modulation, empowering ClusIR to attain remarkable restoration results across a wide range of degradations. Extensive experiments on diverse benchmarks validate that ClusIR reaches competitive performance under several scenarios.

Yichu Xu, Di Wang, Hongzan Jiao et al.

Mamba-based models have recently demonstrated significant potential in hyperspectral image (HSI) classification, primarily due to their ability to perform contextual modeling with linear computational complexity. However, existing Mamba-based approaches often overlook the directional modeling heterogeneity across different land-cover types, leading to limited classification performance. To address these limitations, we propose MambaMoE, a novel spectral-spatial Mixture-of-Experts (MoE) framework, which represents the first MoE-based approach in the HSI classification domain. Specifically, we design a Mixture of Mamba Expert Block (MoMEB) that performs adaptive spectral-spatial feature modeling via a sparse expert activation mechanism. Additionally, we introduce an uncertainty-guided corrective learning (UGCL) strategy that encourages the model to focus on complex regions prone to prediction ambiguity. This strategy dynamically samples supervision signals from regions with high predictive uncertainty, guiding the model to adaptively refine feature representations and thereby enhancing its focus on challenging areas. Extensive experiments conducted on multiple public HSI benchmark datasets show that MambaMoE achieves state-of-the-art performance in both classification accuracy and computational efficiency compared to existing advanced methods, particularly Mamba-based ones. The code will be available online at https://github.com/YichuXu/MambaMoE.

Lu Yang, Jiajia Li, En Ci et al.

Universal Information Extraction (UIE) has garnered significant attention due to its ability to address model explosion problems effectively. Extractive UIE can achieve strong performance using a relatively small model, making it widely adopted. Extractive UIEs generally rely on task instructions for different tasks, including single-target instructions and multiple-target instructions. Single-target instruction UIE enables the extraction of only one type of relation at a time, limiting its ability to model correlations between relations and thus restricting its capability to extract complex relations. While multiple-target instruction UIE allows for the extraction of multiple relations simultaneously, the inclusion of irrelevant relations introduces decision complexity and impacts extraction accuracy. Therefore, for multi-relation extraction, we propose LDNet, which incorporates multi-aspect relation modeling and a label drop mechanism. By assigning different relations to different levels for understanding and decision-making, we reduce decision confusion. Additionally, the label drop mechanism effectively mitigates the impact of irrelevant relations. Experiments show that LDNet outperforms or achieves competitive performance with state-of-the-art systems on 9 tasks, 33 datasets, in both single-modal and multi-modal, few-shot and zero-shot settings.\footnote{https://github.com/Lu-Yang666/LDNet}

Shi Chen, Lefei Zhang, Liangpei Zhang

Mamba has demonstrated exceptional performance in visual tasks due to its powerful global modeling capabilities and linear computational complexity, offering considerable potential in hyperspectral image super-resolution (HSISR). However, in HSISR, Mamba faces challenges as transforming images into 1D sequences neglects the spatial-spectral structural relationships between locally adjacent pixels, and its performance is highly sensitive to input order, which affects the restoration of both spatial and spectral details. In this paper, we propose HSRMamba, a contextual spatial-spectral modeling state space model for HSISR, to address these issues both locally and globally. Specifically, a local spatial-spectral partitioning mechanism is designed to establish patch-wise causal relationships among adjacent pixels in 3D features, mitigating the local forgetting issue. Furthermore, a global spectral reordering strategy based on spectral similarity is employed to enhance the causal representation of similar pixels across both spatial and spectral dimensions. Finally, experimental results demonstrate our HSRMamba outperforms the state-of-the-art methods in quantitative quality and visual results. Code is available at: https://github.com/Tomchenshi/HSRMamba.

Mingni Tang, Jiajia Li, Lu Yang et al.

Symbolic music is represented in two distinct forms: two-dimensional, visually intuitive score images, and one-dimensional, standardized text annotation sequences. While large language models have shown extraordinary potential in music, current research has primarily focused on unimodal symbol sequence text. Existing general-domain visual language models still lack the ability of music notation understanding. Recognizing this gap, we propose NOTA, the first large-scale comprehensive multimodal music notation dataset. It consists of 1,019,237 records, from 3 regions of the world, and contains 3 tasks. Based on the dataset, we trained NotaGPT, a music notation visual large language model. Specifically, we involve a pre-alignment training phase for cross-modal alignment between the musical notes depicted in music score images and their textual representation in ABC notation. Subsequent training phases focus on foundational music information extraction, followed by training on music notation analysis. Experimental results demonstrate that our NotaGPT-7B achieves significant improvement on music understanding, showcasing the effectiveness of NOTA and the training pipeline. Our datasets are open-sourced at https://huggingface.co/datasets/MYTH-Lab/NOTA-dataset.

Xiaokang Zhang, Xuran Xiong, Jianzhong Huang et al.

Remote sensing image segmentation (RSIS) in federated environments has gained increasing attention because it enables collaborative model training across distributed datasets without sharing raw imagery or annotations. Federated RSIS combined with parameter-efficient fine-tuning (PEFT) can unleash the generalization power of pretrained foundation models for real-world applications, with minimal parameter aggregation and communication overhead. However, the dynamic adaptation of pretrained models to heterogeneous client data inevitably increases update uncertainty and compromises the reliability of collaborative optimization due to the lack of uncertainty estimation for each local model. To bridge this gap, we present FedEU, a federated optimization framework for fine-tuning RSIS models driven by evidential uncertainty. Specifically, personalized evidential uncertainty modeling is introduced to quantify epistemic variations of local models and identify high-risk areas under local data distributions. Furthermore, the client-specific feature embedding (CFE) is exploited to enhance channel-aware feature representation while preserving client-specific properties through personalized attention and an element-aware parameter update approach. These uncertainty estimates are uploaded to the server to enable adaptive global aggregation via a Top-k uncertainty-guided weighting (TUW) strategy, which mitigates the impact of distribution shifts and unreliable updates. Extensive experiments on three large-scale heterogeneous datasets demonstrate the superior performance of FedEU. More importantly, FedEU enables balanced model adaptation across diverse clients by explicitly reducing prediction uncertainty, resulting in more robust and reliable federated outcomes. The source codes will be available at https://github.com/zxk688/FedEU.

Xiaokang Zhang, Bo Li, Chufeng Zhou et al.

Pretraining and fine-tuning have emerged as a new paradigm in remote sensing image interpretation. Among them, Masked Autoencoder (MAE)-based pretraining stands out for its strong capability to learn general feature representations via reconstructing masked image regions. However, applying MAE to multispectral remote sensing images remains challenging due to complex backgrounds, indistinct targets, and the lack of semantic guidance during masking, which hinders the learning of underlying structures and meaningful spatial-spectral features. To address this, we propose a simple yet effective approach, Spectral Index-Guided MAE (SIGMAE), for multispectral image pretraining. The core idea is to incorporate domain-specific spectral indices as prior knowledge to guide dynamic token masking toward informative regions. SIGMAE introduces Semantic Saliency-Guided Dynamic Token Masking (SSDTM), a curriculum-style strategy that quantifies each patch's semantic richness and internal heterogeneity to adaptively select the most informative tokens during training. By prioritizing semantically salient regions and progressively increasing sample difficulty, SSDTM enhances spectrally rich and structurally aware representation learning, mitigates overfitting, and reduces redundant computation compared with random masking. Extensive experiments on five widely used datasets covering various downstream tasks, including scene classification, semantic segmentation, object extraction and change detection, demonstrate that SIGMAE outperforms other pretrained geospatial foundation models. Moreover, it exhibits strong spatial-spectral reconstruction capability, even with a 90% mask ratio, and improves complex target recognition under limited labeled data. The source codes and model weights will be released at https://github.com/zxk688/SIGMAE.

Shengkai Hu, Haozhe Qi, Jun Wan et al.

Recent advances in deep learning have significantly improved facial landmark detection. However, existing facial landmark detection datasets often define different numbers of landmarks, and most mainstream methods can only be trained on a single dataset. This limits the model generalization to different datasets and hinders the development of a unified model. To address this issue, we propose Proto-Former, a unified, adaptive, end-to-end facial landmark detection framework that explicitly enhances dataset-specific facial structural representations (i.e., prototype). Proto-Former overcomes the limitations of single-dataset training by enabling joint training across multiple datasets within a unified architecture. Specifically, Proto-Former comprises two key components: an Adaptive Prototype-Aware Encoder (APAE) that performs adaptive feature extraction and learns prototype representations, and a Progressive Prototype-Aware Decoder (PPAD) that refines these prototypes to generate prompts that guide the model's attention to key facial regions. Furthermore, we introduce a novel Prototype-Aware (PA) loss, which achieves optimal path finding by constraining the selection weights of prototype experts. This loss function effectively resolves the problem of prototype expert addressing instability during multi-dataset training, alleviates gradient conflicts, and enables the extraction of more accurate facial structure features. Extensive experiments on widely used benchmark datasets demonstrate that our Proto-Former achieves superior performance compared to existing state-of-the-art methods. The code is publicly available at: https://github.com/Husk021118/Proto-Former.

Teng Xiao, Zuchao Li, Lefei Zhang

Recent advances in multimodal large language models (LLMs) have led to significant progress in understanding, generation, and retrieval tasks. However, current solutions often treat these tasks in isolation or require training LLMs from scratch, resulting in high computational costs and limited generalization across modalities. In this work, we present OmniBridge, a unified and modular multimodal framework that supports vision-language understanding, generation, and retrieval within a unified architecture. OmniBridge adopts a language-centric design that reuses pretrained LLMs and introduces a lightweight bidirectional latent alignment module. To address the challenge of task interference, we propose a two-stage decoupled training strategy: supervised fine-tuning and latent space alignment for aligning LLM behavior with multimodal reasoning, and semantic-guided diffusion training to align cross-modal latent spaces via learnable query embeddings. Extensive experiments across a wide range of benchmarks demonstrate that OmniBridge achieves competitive or state-of-the-art performance in all three tasks. Moreover, our results highlight the effectiveness of latent space alignment for unifying multimodal modeling under a shared representation space. Code and models are released at https://github.com/xiao-xt/OmniBridge.

Liang Lv, Di Wang, Jing Zhang et al.

Semi-supervised semantic segmentation (S4) has advanced remote sensing (RS) analysis by leveraging unlabeled data through pseudo-labeling and consistency learning. However, existing S4 studies often rely on small-scale datasets and models, limiting their practical applicability. To address this, we propose S5, the first scalable framework for semi-supervised semantic segmentation in RS, which unlocks the potential of vast unlabeled Earth observation data typically underutilized due to costly pixel-level annotations. Built upon existing large-scale RS datasets, S5 introduces a data selection strategy that integrates entropy-based filtering and diversity expansion, resulting in the RS4P-1M dataset. Using this dataset, we systematically scale up S4 methods by pre-training RS foundation models (RSFMs) of varying sizes on this extensive corpus, significantly boosting their performance on land cover segmentation and object detection tasks. Furthermore, during fine-tuning, we incorporate a Mixture-of-Experts (MoE)-based multi-dataset fine-tuning approach, which enables efficient adaptation to multiple RS benchmarks with fewer parameters. This approach improves the generalization and versatility of RSFMs across diverse RS benchmarks. The resulting RSFMs achieve state-of-the-art performance across all benchmarks, underscoring the viability of scaling semi-supervised learning for RS applications. All datasets, code, and models will be released at https://github.com/MiliLab/S5

Zihong Zhang, Liqi He, Zuchao Li et al.

Word segmentation stands as a cornerstone of Natural Language Processing (NLP). Based on the concept of "comprehend first, segment later", we propose a new framework to explore the limit of unsupervised word segmentation with Large Language Models (LLMs) and evaluate the semantic understanding capabilities of LLMs based on word segmentation. We employ current mainstream LLMs to perform word segmentation across multiple languages to assess LLMs' "comprehension". Our findings reveal that LLMs are capable of following simple prompts to segment raw text into words. There is a trend suggesting that models with more parameters tend to perform better on multiple languages. Additionally, we introduce a novel unsupervised method, termed LLACA ($\textbf{L}$arge $\textbf{L}$anguage Model-Inspired $\textbf{A}$ho-$\textbf{C}$orasick $\textbf{A}$utomaton). Leveraging the advanced pattern recognition capabilities of Aho-Corasick automata, LLACA innovatively combines these with the deep insights of well-pretrained LLMs. This approach not only enables the construction of a dynamic $n$-gram model that adjusts based on contextual information but also integrates the nuanced understanding of LLMs, offering significant improvements over traditional methods. Our source code is available at https://github.com/hkr04/LLACA

Di Wang, Meiqi Hu, Yao Jin et al.

Accurate hyperspectral image (HSI) interpretation is critical for providing valuable insights into various earth observation-related applications such as urban planning, precision agriculture, and environmental monitoring. However, existing HSI processing methods are predominantly task-specific and scene-dependent, which severely limits their ability to transfer knowledge across tasks and scenes, thereby reducing the practicality in real-world applications. To address these challenges, we present HyperSIGMA, a vision transformer-based foundation model that unifies HSI interpretation across tasks and scenes, scalable to over one billion parameters. To overcome the spectral and spatial redundancy inherent in HSIs, we introduce a novel sparse sampling attention (SSA) mechanism, which effectively promotes the learning of diverse contextual features and serves as the basic block of HyperSIGMA. HyperSIGMA integrates spatial and spectral features using a specially designed spectral enhancement module. In addition, we construct a large-scale hyperspectral dataset, HyperGlobal-450K, for pre-training, which contains about 450K hyperspectral images, significantly surpassing existing datasets in scale. Extensive experiments on various high-level and low-level HSI tasks demonstrate HyperSIGMA's versatility and superior representational capability compared to current state-of-the-art methods. Moreover, HyperSIGMA shows significant advantages in scalability, robustness, cross-modal transferring capability, real-world applicability, and computational efficiency. The code and models will be released at https://github.com/WHU-Sigma/HyperSIGMA.

Cong Yang, Zuchao Li, Lefei Zhang

Recently, large multimodal models have built a bridge from visual to textual information, but they tend to underperform in remote sensing scenarios. This underperformance is due to the complex distribution of objects and the significant scale differences among targets in remote sensing images, leading to visual ambiguities and insufficient descriptions by these multimodal models. Moreover, the lack of multimodal fine-tuning data specific to the remote sensing field makes it challenging for the model's behavior to align with user queries. To address these issues, this paper proposes an attribute-guided \textbf{Multi-Granularity Instruction Multimodal Model (MGIMM)} for remote sensing image detailed description. MGIMM guides the multimodal model to learn the consistency between visual regions and corresponding text attributes (such as object names, colors, and shapes) through region-level instruction tuning. Then, with the multimodal model aligned on region-attribute, guided by multi-grain visual features, MGIMM fully perceives both region-level and global image information, utilizing large language models for comprehensive descriptions of remote sensing images. Due to the lack of a standard benchmark for generating detailed descriptions of remote sensing images, we construct a dataset featuring 38,320 region-attribute pairs and 23,463 image-detailed description pairs. Compared with various advanced methods on this dataset, the results demonstrate the effectiveness of MGIMM's region-attribute guided learning approach. Code can be available at https://github.com/yangcong356/MGIMM.git