Yilin Wang

Pavan C. Madhusudana, Neil Birkbeck, Yilin Wang et al.

Perceptual video quality assessment (VQA) is an integral component of many streaming and video sharing platforms. Here we consider the problem of learning perceptually relevant video quality representations in a self-supervised manner. Distortion type identification and degradation level determination is employed as an auxiliary task to train a deep learning model containing a deep Convolutional Neural Network (CNN) that extracts spatial features, as well as a recurrent unit that captures temporal information. The model is trained using a contrastive loss and we therefore refer to this training framework and resulting model as CONtrastive VIdeo Quality EstimaTor (CONVIQT). During testing, the weights of the trained model are frozen, and a linear regressor maps the learned features to quality scores in a no-reference (NR) setting. We conduct comprehensive evaluations of the proposed model on multiple VQA databases by analyzing the correlations between model predictions and ground-truth quality ratings, and achieve competitive performance when compared to state-of-the-art NR-VQA models, even though it is not trained on those databases. Our ablation experiments demonstrate that the learned representations are highly robust and generalize well across synthetic and realistic distortions. Our results indicate that compelling representations with perceptual bearing can be obtained using self-supervised learning. The implementations used in this work have been made available at https://github.com/pavancm/CONVIQT.

Yash Savani, Branislav Kveton, Yuchen Liu et al. · cmu, stanford

Flow-GRPO successfully applies reinforcement learning to flow models, but uses uniform credit assignment across all steps. This ignores the temporal structure of diffusion generation: early steps determine composition and content (low-frequency structure), while late steps resolve details and textures (high-frequency details). Moreover, assigning uniform credit based solely on the final image can inadvertently reward suboptimal intermediate steps, especially when errors are corrected later in the diffusion trajectory. We propose Stepwise-Flow-GRPO, which assigns credit based on each step's reward improvement. By leveraging Tweedie's formula to obtain intermediate reward estimates and introducing gain-based advantages, our method achieves superior sample efficiency and faster convergence. We also introduce a DDIM-inspired SDE that improves reward quality while preserving stochasticity for policy gradients.

Cheng-Han Lee, Maniratnam Mandal, Neil Birkbeck et al.

With the rise of mobile video consumption on diverse handheld display resolutions and orientation modes, altering videos to aspect ratios poses challenges. Static cropping and border padding often compromises visual quality, while warping may distort a video's intended meaning. Here we advocate for a more effective approach: cropping significant regions within video frames in a temporal manner, while minimizing distortion and preserving essential content. One barrier to solving this problem is the lack of sufficiently large-scale database devoted to informing these tasks. Towards filling this gap, we introduce the LIVE-YouTube Video Cropping (LIVE-YT VC) database, featuring 1800 videos, annotated by 90 human subjects. Using videos sourced from the YouTube-UGC and LSVQ Databases, this new resource is the largest publicly-available subjective video portrait region cropping database. We also introduce a post-processed version of the database, called LIVE-YT VC++, whereby a novel intra-frame temporal filter was deployed to smooth subjective annotations within each video. We demonstrate the usefulness of this new data resource using the SmartVidCrop algorithm and state-of-the-art video grounding models, in hopes of establishing our subjective dataset as a benchmark for future research. Our contributions offer a resource for advancing video aspect ratio transformation models towards ensuring that reshaped mobile-friendly video content retains its quality and meaning. Since our labels bear resemblances to video saliency annotations, we also conducted an additional analysis to explore the similarity between our labels and video saliency predictions. Finally, we repurposed state-of-the-art video grounding models for aspect ratio change tasks, and fine-tuned them on our dataset. As a service to the research community, we plan to open source the project.

Xiangxu Yu, Zhenqiang Ying, Neil Birkbeck et al.

The rising popularity of online User-Generated-Content (UGC) in the form of streamed and shared videos, has hastened the development of perceptual Video Quality Assessment (VQA) models, which can be used to help optimize their delivery. Gaming videos, which are a relatively new type of UGC videos, are created when skilled gamers post videos of their gameplay. These kinds of screenshots of UGC gameplay videos have become extremely popular on major streaming platforms like YouTube and Twitch. Synthetically-generated gaming content presents challenges to existing VQA algorithms, including those based on natural scene/video statistics models. Synthetically generated gaming content presents different statistical behavior than naturalistic videos. A number of studies have been directed towards understanding the perceptual characteristics of professionally generated gaming videos arising in gaming video streaming, online gaming, and cloud gaming. However, little work has been done on understanding the quality of UGC gaming videos, and how it can be characterized and predicted. Towards boosting the progress of gaming video VQA model development, we conducted a comprehensive study of subjective and objective VQA models on UGC gaming videos. To do this, we created a novel UGC gaming video resource, called the LIVE-YouTube Gaming video quality (LIVE-YT-Gaming) database, comprised of 600 real UGC gaming videos. We conducted a subjective human study on this data, yielding 18,600 human quality ratings recorded by 61 human subjects. We also evaluated a number of state-of-the-art (SOTA) VQA models on the new database, including a new one, called GAME-VQP, based on both natural video statistics and CNN-learned features. To help support work in this field, we are making the new LIVE-YT-Gaming Database, publicly available through the link: https://live.ece.utexas.edu/research/LIVE-YT-Gaming/index.html .

Jeya Maria Jose Valanarasu, He Zhang, Jianming Zhang et al.

Current image harmonization methods consider the entire background as the guidance for harmonization. However, this may limit the capability for user to choose any specific object/person in the background to guide the harmonization. To enable flexible interaction between user and harmonization, we introduce interactive harmonization, a new setting where the harmonization is performed with respect to a selected \emph{region} in the reference image instead of the entire background. A new flexible framework that allows users to pick certain regions of the background image and use it to guide the harmonization is proposed. Inspired by professional portrait harmonization users, we also introduce a new luminance matching loss to optimally match the color/luminance conditions between the composite foreground and select reference region. This framework provides more control to the image harmonization pipeline achieving visually pleasing portrait edits. Furthermore, we also introduce a new dataset carefully curated for validating portrait harmonization. Extensive experiments on both synthetic and real-world datasets show that the proposed approach is efficient and robust compared to previous harmonization baselines, especially for portraits. Project Webpage at \href{https://jeya-maria-jose.github.io/IPH-web/}{https://jeya-maria-jose.github.io/IPH-web/}

Junjie Ke, Tianhao Zhang, Yilin Wang et al.

No-reference video quality assessment (NR-VQA) for user generated content (UGC) is crucial for understanding and improving visual experience. Unlike video recognition tasks, VQA tasks are sensitive to changes in input resolution. Since large amounts of UGC videos nowadays are 720p or above, the fixed and relatively small input used in conventional NR-VQA methods results in missing high-frequency details for many videos. In this paper, we propose a novel Transformer-based NR-VQA framework that preserves the high-resolution quality information. With the multi-resolution input representation and a novel multi-resolution patch sampling mechanism, our method enables a comprehensive view of both the global video composition and local high-resolution details. The proposed approach can effectively aggregate quality information across different granularities in spatial and temporal dimensions, making the model robust to input resolution variations. Our method achieves state-of-the-art performance on large-scale UGC VQA datasets LSVQ and LSVQ-1080p, and on KoNViD-1k and LIVE-VQC without fine-tuning.

Pavan C. Madhusudana, Neil Birkbeck, Yilin Wang et al.

We consider the problem of capturing distortions arising from changes in frame rate as part of Video Quality Assessment (VQA). Variable frame rate (VFR) videos have become much more common, and streamed videos commonly range from 30 frames per second (fps) up to 120 fps. VFR-VQA offers unique challenges in terms of distortion types as well as in making non-uniform comparisons of reference and distorted videos having different frame rates. The majority of current VQA models require compared videos to be of the same frame rate, but are unable to adequately account for frame rate artifacts. The recently proposed Generalized Entropic Difference (GREED) VQA model succeeds at this task, using natural video statistics models of entropic differences of temporal band-pass coefficients, delivering superior performance on predicting video quality changes arising from frame rate distortions. Here we propose a simple fusion framework, whereby temporal features from GREED are combined with existing VQA models, towards improving model sensitivity towards frame rate distortions. We find through extensive experiments that this feature fusion significantly boosts model performance on both HFR/VFR datasets as well as fixed frame rate (FFR) VQA databases. Our results suggest that employing efficient temporal representations can result much more robust and accurate VQA models when frame rate variations can occur.

Yilin Wang, Yifei Yu, Kong Sun et al.

In industry, the reliability of rotating machinery is critical for production efficiency and safety. Current methods of Prognostics and Health Management (PHM) often rely on task-specific models, which face significant challenges in handling diverse datasets with varying signal characteristics, fault modes and operating conditions. Inspired by advancements in generative pretrained models, we propose RmGPT, a unified model for diagnosis and prognosis tasks. RmGPT introduces a novel generative token-based framework, incorporating Signal Tokens, Prompt Tokens, Time-Frequency Task Tokens and Fault Tokens to handle heterogeneous data within a unified model architecture. We leverage self-supervised learning for robust feature extraction and introduce a next signal token prediction pretraining strategy, alongside efficient prompt learning for task-specific adaptation. Extensive experiments demonstrate that RmGPT significantly outperforms state-of-the-art algorithms, achieving near-perfect accuracy in diagnosis tasks and exceptionally low errors in prognosis tasks. Notably, RmGPT excels in few-shot learning scenarios, achieving 82\% accuracy in 16-class one-shot experiments, highlighting its adaptability and robustness. This work establishes RmGPT as a powerful PHM foundation model for rotating machinery, advancing the scalability and generalizability of PHM solutions. \textbf{Code is available at: https://github.com/Pandalin98/RmGPT.

Renyu Yang, Jian Jin, Lili Meng et al.

Audio-visual quality assessment (AVQA) research has been stalled by limitations of existing datasets: they are typically small in scale, with insufficient diversity in content and quality, and annotated only with overall scores. These shortcomings provide limited support for model development and multimodal perception research. We propose a practical approach for AVQA dataset construction. First, we design a crowdsourced subjective experiment framework for AVQA, breaks the constraints of in-lab settings and achieves reliable annotation across varied environments. Second, a systematic data preparation strategy is further employed to ensure broad coverage of both quality levels and semantic scenarios. Third, we extend the dataset with additional annotations, enabling research on multimodal perception mechanisms and their relation to content. Finally, we validate this approach through YT-NTU-AVQ, the largest and most diverse AVQA dataset to date, consisting of 1,620 user-generated audio and video (A/V) sequences. The dataset and platform code are available at https://github.com/renyu12/YT-NTU-AVQ

Yuanpei Zhao, Jie Lin, Chao Zhang et al.

Pairwise preferences and pointwise ratings are the two dominant annotation protocols in image aesthetic assessment (IAA), yet existing benchmarks adopt only one, leaving their complementarity unmeasured under controlled conditions. We introduce PPaint, a matched dual-protocol benchmark in which 15 domain experts, 5 per category, annotate 150 Chinese paintings under both protocols across five aesthetic dimensions, collecting 45,900 pairwise expert judgments through a locally dense preference design alongside the matched ratings. The matched design reveals complementary strengths: preferences yield more consistent ordinal rankings, while ratings anchor the absolute score scale. Fusing both signals via two independent preference-to-score methods yields a fused expert ground truth on which the two constructions converge to nearly identical scores. The same preference-to-score principle extends to label-free VLM training. PSDistill converts VLM pairwise judgments into calibrated pseudo-scores via an Elo reference pool, and trains the same VLM with confidence-weighted ranking optimization to produce a single-pass aesthetic scorer. Trained on a single painting category, the distilled Qwen3-VL-8B improves mean SRCC from 0.504 to 0.709 across all three categories, outperforming all open-source baselines including the dedicated aesthetic model ArtiMuse and matching closed-source Gemini-3.1-Pro within 0.04 SRCC at single-pass inference cost, with cross-domain transfer further validated on APDDv2. We will release the full PPaint dataset and training code.

Shreshth Saini, Avinab Saha, Balu Adsumilli et al.

Masked Diffusion Models (MDMs) have emerged as a promising non-autoregressive paradigm for generative tasks, offering parallel decoding and bidirectional context utilization. However, current sampling methods rely on simple confidence-based heuristics that ignore the long-term impact of local decisions, leading to trajectory lock-in where early hallucinations cascade into global incoherence. While search-based methods mitigate this, they incur prohibitive computational costs ($O(K)$ forward passes per step). In this work, we propose Backward-on-Entropy (BoE) Steering, a gradient-guided inference framework that approximates infinite-horizon lookahead via a single backward pass. We formally derive the Token Influence Score (TIS) from a first-order expansion of the trajectory cost functional, proving that the gradient of future entropy with respect to input embeddings serves as an optimal control signal for minimizing uncertainty. To ensure scalability, we introduce \texttt{ActiveQueryAttention}, a sparse adjoint primitive that exploits the structure of the masking objective to reduce backward pass complexity. BoE achieves a superior Pareto frontier for inference-time scaling compared to existing unmasking methods, demonstrating that gradient-guided steering offers a mathematically principled and efficient path to robust non-autoregressive generation. We will release the code.

Yuxuan Mu, Xinxin Zuo, Chuan Guo et al.

We present GSD, a diffusion model approach based on Gaussian Splatting (GS) representation for 3D object reconstruction from a single view. Prior works suffer from inconsistent 3D geometry or mediocre rendering quality due to improper representations. We take a step towards resolving these shortcomings by utilizing the recent state-of-the-art 3D explicit representation, Gaussian Splatting, and an unconditional diffusion model. This model learns to generate 3D objects represented by sets of GS ellipsoids. With these strong generative 3D priors, though learning unconditionally, the diffusion model is ready for view-guided reconstruction without further model fine-tuning. This is achieved by propagating fine-grained 2D features through the efficient yet flexible splatting function and the guided denoising sampling process. In addition, a 2D diffusion model is further employed to enhance rendering fidelity, and improve reconstructed GS quality by polishing and re-using the rendered images. The final reconstructed objects explicitly come with high-quality 3D structure and texture, and can be efficiently rendered in arbitrary views. Experiments on the challenging real-world CO3D dataset demonstrate the superiority of our approach. Project page: https://yxmu.foo/GSD/

Yilin Wang, Yiheng Feng

Model-based and learning-based methods are two major types of methodologies to model car following behaviors. Model-based methods describe the car-following behaviors with explicit mathematical equations, while learning-based methods focus on getting a mapping between inputs and outputs. Both types of methods have advantages and weaknesses. Meanwhile, most car-following models are generative and only consider the inputs of the speed, position, and acceleration of the last time step. To address these issues, this study proposes a novel framework called IDM-Follower that can generate a sequence of following vehicle trajectory by a recurrent autoencoder informed by a physical car-following model, the Intelligent Driving Model (IDM).We implement a novel structure with two independent encoders and a self-attention decoder that could sequentially predict the following trajectories. A loss function considering the discrepancies between predictions and labeled data integrated with discrepancies from model-based predictions is implemented to update the neural network parameters. Numerical experiments with multiple settings on simulation and NGSIM datasets show that the IDM-Follower can improve the prediction performance compared to the model-based or learning-based methods alone. Analysis on different noise levels also shows good robustness of the model.

Xiangyu Huang, Caidan Zhao, Yilin Wang et al.

Video anomaly detection refers to the identification of events that deviate from the expected behavior. Due to the lack of anomalous samples in training, video anomaly detection becomes a very challenging task. Existing methods almost follow a reconstruction or future frame prediction mode. However, these methods ignore the consistency between appearance and motion information of samples, which limits their anomaly detection performance. Anomalies only occur in the moving foreground of surveillance videos, so the semantics expressed by video frame sequences and optical flow without background information in anomaly detection should be highly consistent and significant for anomaly detection. Based on this idea, we propose Appearance-Motion Semantics Representation Consistency (AMSRC), a framework that uses normal data's appearance and motion semantic representation consistency to handle anomaly detection. Firstly, we design a two-stream encoder to encode the appearance and motion information representations of normal samples and introduce constraints to further enhance the consistency of the feature semantics between appearance and motion information of normal samples so that abnormal samples with low consistency appearance and motion feature representation can be identified. Moreover, the lower consistency of appearance and motion features of anomalous samples can be used to generate predicted frames with larger reconstruction error, which makes anomalies easier to spot. Experimental results demonstrate the effectiveness of the proposed method.

Yilin Wang, Zeyuan Chen, Liangjun Zhong et al.

In this paper, we introduce a novel generative model, Diffusion Layout Transformers without Autoencoder (Dolfin), which significantly improves the modeling capability with reduced complexity compared to existing methods. Dolfin employs a Transformer-based diffusion process to model layout generation. In addition to an efficient bi-directional (non-causal joint) sequence representation, we further propose an autoregressive diffusion model (Dolfin-AR) that is especially adept at capturing rich semantic correlations for the neighboring objects, such as alignment, size, and overlap. When evaluated against standard generative layout benchmarks, Dolfin notably improves performance across various metrics (fid, alignment, overlap, MaxIoU and DocSim scores), enhancing transparency and interoperability in the process. Moreover, Dolfin's applications extend beyond layout generation, making it suitable for modeling geometric structures, such as line segments. Our experiments present both qualitative and quantitative results to demonstrate the advantages of Dolfin.

Yilin Wang, Farzan Farnia

Black-box adversarial attacks designing adversarial examples for unseen neural networks (NNs) have received great attention over the past years. While several successful black-box attack schemes have been proposed in the literature, the underlying factors driving the transferability of black-box adversarial examples still lack a thorough understanding. In this paper, we aim to demonstrate the role of the generalization properties of the substitute classifier used for generating adversarial examples in the transferability of the attack scheme to unobserved NN classifiers. To do this, we apply the max-min adversarial example game framework and show the importance of the generalization properties of the substitute NN in the success of the black-box attack scheme in application to different NN classifiers. We prove theoretical generalization bounds on the difference between the attack transferability rates on training and test samples. Our bounds suggest that a substitute NN with better generalization behavior could result in more transferable adversarial examples. In addition, we show that standard operator norm-based regularization methods could improve the transferability of the designed adversarial examples. We support our theoretical results by performing several numerical experiments showing the role of the substitute network's generalization in generating transferable adversarial examples. Our empirical results indicate the power of Lipschitz regularization methods in improving the transferability of adversarial examples.

Jiayi Liu, Yilin Wang, Michael Beyeler

Cloud-based machine learning is increasingly explored as a preprocessing strategy for next-generation visual neuroprostheses, where advanced scene understanding may exceed the computational and energy constraints of battery-powered visual processing units. Offloading computation to remote servers enables the use of state-of-the-art vision models, but also introduces sensitivity to network latency, jitter, and packet loss, which can disrupt the temporal consistency of the delivered neural stimulus. In this work, we examine the feasibility of cloud-assisted visual preprocessing for artificial vision by framing remote inference as a perceptually constrained systems problem. We present a network-adaptive cloud-assisted pipeline in which real-time round-trip-time feedback is used to dynamically modulate image resolution, compression, and transmission rate, explicitly prioritizing temporal continuity under adverse network conditions. PIDNet is used as a fixed real-time semantic segmentation backbone, allowing us to isolate how network-adaptive input encoding affects communication delay, inference time, and perceptual fidelity. Results show that adaptive visual encoding substantially reduces end-to-end latency during network congestion, with only modest degradation of global scene structure, while boundary precision degrades more sharply. Together, these findings delineate operating regimes in which cloud-assisted preprocessing may remain viable for future visual neuroprostheses and underscore the importance of network-aware adaptation for maintaining perceptual stability.

Shreshth Saini, Hakan Gedik, Neil Birkbeck et al.

The rapid adoption of HDR-capable devices has created a pressing need to convert the 8-bit Standard Dynamic Range (SDR) content into perceptually and physically accurate 10-bit High Dynamic Range (HDR). Existing inverse tone-mapping (ITM) methods often rely on fixed tone-mapping operators that struggle to generalize to real-world degradations, stylistic variations, and camera pipelines, frequently producing clipped highlights, desaturated colors, or unstable tone reproduction. We introduce LumaFlux, a first physically and perceptually guided diffusion transformer (DiT) for SDR-to-HDR reconstruction by adapting a large pretrained DiT. Our LumaFlux introduces (1) a Physically-Guided Adaptation (PGA) module that injects luminance, spatial descriptors, and frequency cues into attention through low-rank residuals; (2) a Perceptual Cross-Modulation (PCM) layer that stabilizes chroma and texture via FiLM conditioning from vision encoder features; and (3) an HDR Residual Coupler that fuses physical and perceptual signals under a timestep- and layer-adaptive modulation schedule. Finally, a lightweight Rational-Quadratic Spline decoder reconstructs smooth, interpretable tone fields for highlight and exposure expansion, enhancing the output of the VAE decoder to generate HDR. To enable robust HDR learning, we curate the first large-scale SDR-HDR training corpus. For fair and reproducible comparison, we further establish a new evaluation benchmark, comprising HDR references and corresponding expert-graded SDR versions. Across benchmarks, LumaFlux outperforms state-of-the-art baselines, achieving superior luminance reconstruction and perceptual color fidelity with minimal additional parameters.

Yuchun Fan, Bei Li, Peiguang Li et al.

Reinforcement learning has proven effective for enhancing multi-step reasoning in large language models (LLMs), yet its benefits have not fully translated to multilingual contexts. Existing methods struggle with a fundamental trade-off: prioritizing input-language consistency severely hampers reasoning quality, while prioritizing reasoning often leads to unintended language drift toward English. We address this challenge with LANG, a novel framework that leverages language-conditioned hints to guide exploration in non-English reasoning tasks. Our method incorporates two key mechanisms to prevent dependency on these hints: a progressive decay schedule that gradually withdraws scaffolding, and a language-adaptive switch that tailors learning horizons to specific language difficulties. Empirical results on challenging multilingual mathematical benchmarks reveal that LANG substantially enhances reasoning performance without compromising language consistency. Moreover, we show that our framework generalizes beyond mathematics, fostering more consistent language alignment across model layers

Shreshth Saini, Bowen Chen, Neil Birkbeck et al.

High Dynamic Range (HDR) user-generated (UGC) videos are rapidly proliferating across social platforms, yet most perceptual video quality assessment (VQA) systems remain tailored to Standard Dynamic Range (SDR). HDR has a higher bit depth, wide color gamut, and elevated luminance range, exposing distortions such as near-black crushing, highlight clipping, banding, and exposure flicker that amplify UGC artifacts and challenge SDR models. To catalyze progress, we curate Beyond8Bits, a large-scale subjective dataset of 44K videos from 6.5K sources with over 1.5M crowd ratings, spanning diverse scenes, capture conditions, and compression settings. We further introduce HDR-Q, the first Multimodal Large Language Model (MLLM) for HDR-UGC VQA. We propose (i) a novel HDR-aware vision encoder to produce HDR-sensitive embeddings, and (ii) HDR-Aware Policy Optimization (HAPO), an RL finetuning framework that anchors reasoning to HDR cues. HAPO augments GRPO via an HDR-SDR contrastive KL that encourages token reliance on HDR inputs and a Gaussian weighted regression reward for fine-grained MOS calibration. Across Beyond8Bits and public HDR-VQA benchmarks, HDR-Q delivers state-of-the-art performance.

Dongxing Mao, Yilin Wang, Linjie Li et al.

Despite recent advances in text-to-image generation, models still struggle to accurately render prompt-specified text with correct spatial layout -- especially in multi-span, structured settings. This challenge is driven not only by the lack of datasets that align prompts with the exact text and layout expected in the image, but also by the absence of effective metrics for evaluating layout quality. To address these issues, we introduce TextGround4M, a large-scale dataset of over 4 million prompt-image pairs, each annotated with span-level text grounded in the prompt and corresponding bounding boxes. This enables fine-grained supervision for layout-aware, prompt-grounded text rendering. Building on this, we propose a lightweight training strategy for autoregressive T2I models that appends layout-aware span tokens during training, without altering model architecture or inference behavior. We further construct a benchmark with stratified layout complexity to evaluate both open-source and proprietary models in a zero-shot setting. In addition, we introduce two layout-aware metrics to address the long-standing lack of spatial evaluation in text rendering. Our results show that models trained on TextGround4M outperform strong baselines in text fidelity, spatial accuracy, and prompt consistency, highlighting the importance of fine-grained layout supervision for grounded T2I generation.

Yilin Wang, Xinyi Hu, Matthew R. Gormley

Most pretrained language models rely on subword tokenization, which processes text as a sequence of subword tokens. However, different granularities of text, such as characters, subwords, and words, can contain different kinds of information. Previous studies have shown that incorporating multiple input granularities improves model generalization, yet very few of them outputs useful representations for each granularity. In this paper, we introduce the entanglement model, aiming to combine character and subword language models. Inspired by vision-language models, our model treats characters and subwords as separate modalities, and it generates mutually informed representations for both granularities as output. We evaluate our model on text classification, named entity recognition, POS-tagging, and character-level sequence labeling (intraword code-switching). Notably, the entanglement model outperforms its backbone language models, particularly in the presence of noisy texts and low-resource languages. Furthermore, the entanglement model even outperforms larger pre-trained models on all English sequence labeling tasks and classification tasks. We make our code publically available.

Nannan Li, Qing Liu, Krishna Kumar Singh et al.

Human image editing includes tasks like changing a person's pose, their clothing, or editing the image according to a text prompt. However, prior work often tackles these tasks separately, overlooking the benefit of mutual reinforcement from learning them jointly. In this paper, we propose UniHuman, a unified model that addresses multiple facets of human image editing in real-world settings. To enhance the model's generation quality and generalization capacity, we leverage guidance from human visual encoders and introduce a lightweight pose-warping module that can exploit different pose representations, accommodating unseen textures and patterns. Furthermore, to bridge the disparity between existing human editing benchmarks with real-world data, we curated 400K high-quality human image-text pairs for training and collected 2K human images for out-of-domain testing, both encompassing diverse clothing styles, backgrounds, and age groups. Experiments on both in-domain and out-of-domain test sets demonstrate that UniHuman outperforms task-specific models by a significant margin. In user studies, UniHuman is preferred by the users in an average of 77% of cases. Our project is available at https://github.com/NannanLi999/UniHuman.

Weidong Bao, Yilin Wang, Ruyu Gao et al.

Knowledge Graphs (KGs) are foundational to applications such as search, question answering, and recommendation. Conventional knowledge graph construction methods are predominantly static, rely ing on a single-step construction from a fixed corpus with a prede f ined schema. However, such methods are suboptimal for real-world sce narios where data arrives dynamically, as incorporating new informa tion requires complete and computationally expensive graph reconstruc tions. Furthermore, predefined schemas hinder the flexibility of knowl edge graph construction. To address these limitations, we introduce DIAL KG, a closed-loop framework for incremental KG construction orches trated by a Meta-Knowledge Base (MKB). The framework oper ates in a three-stage cycle: (i) Dual-Track Extraction, which ensures knowledge completeness by defaulting to triple generation and switching to event extraction for complex knowledge; (ii) Governance Adjudica tion, which ensures the fidelity and currency of extracted facts to prevent hallucinations and knowledge staleness; and (iii) Schema Evolution, in which new schemas are induced from validated knowledge to guide subsequent construction cycles, and knowledge from the current round is incrementally applied to the existing KG. Extensive experiments demon strate that our framework achieves state-of-the-art (SOTA) performance in the quality of both the constructed graph and the induced schemas.

Jiongze Yu, Xiangbo Gao, Pooja Verlani et al.

Video Super-Resolution (VSR) aims to restore high-quality video frames from low-resolution (LR) estimates, yet most existing VSR approaches behave like black boxes at inference time: users cannot reliably correct unexpected artifacts, but instead can only accept whatever the model produces. In this paper, we propose a novel interactive VSR framework dubbed SparkVSR that makes sparse keyframes a simple and expressive control signal. Specifically, users can first super-resolve or optionally a small set of keyframes using any off-the-shelf image super-resolution (ISR) model, then SparkVSR propagates the keyframe priors to the entire video sequence while remaining grounded by the original LR video motion. Concretely, we introduce a keyframe-conditioned latent-pixel two-stage training pipeline that fuses LR video latents with sparsely encoded HR keyframe latents to learn robust cross-space propagation and refine perceptual details. At inference time, SparkVSR supports flexible keyframe selection (manual specification, codec I-frame extraction, or random sampling) and a reference-free guidance mechanism that continuously balances keyframe adherence and blind restoration, ensuring robust performance even when reference keyframes are absent or imperfect. Experiments on multiple VSR benchmarks demonstrate improved temporal consistency and strong restoration quality, surpassing baselines by up to 24.6%, 21.8%, and 5.6% on CLIP-IQA, DOVER, and MUSIQ, respectively, enabling controllable, keyframe-driven video super-resolution. Moreover, we demonstrate that SparkVSR is a generic interactive, keyframe-conditioned video processing framework as it can be applied out of the box to unseen tasks such as old-film restoration and video style transfer. Our project page is available at: https://sparkvsr.github.io/

Zhengmeng Xu, Yujie Wang, Xiaotong Feng et al.

We propose a time series forecasting method named Quantum Gramian Angular Field (QGAF). This approach merges the advantages of quantum computing technology with deep learning, aiming to enhance the precision of time series classification and forecasting. We successfully transformed stock return time series data into two-dimensional images suitable for Convolutional Neural Network (CNN) training by designing specific quantum circuits. Distinct from the classical Gramian Angular Field (GAF) approach, QGAF's uniqueness lies in eliminating the need for data normalization and inverse cosine calculations, simplifying the transformation process from time series data to two-dimensional images. To validate the effectiveness of this method, we conducted experiments on datasets from three major stock markets: the China A-share market, the Hong Kong stock market, and the US stock market. Experimental results revealed that compared to the classical GAF method, the QGAF approach significantly improved time series prediction accuracy, reducing prediction errors by an average of 25% for Mean Absolute Error (MAE) and 48% for Mean Squared Error (MSE). This research confirms the potential and promising prospects of integrating quantum computing with deep learning techniques in financial time series forecasting.

Yilin Wang, Yuchun Fan, Jiaoyang Li et al.

Retrieval-augmented generation (RAG) systems have made significant progress in solving complex multi-hop question answering (QA) tasks in the English scenario. However, RAG systems inevitably face the application scenario of retrieving across multilingual corpora and queries, leaving several open challenges. The first one involves the absence of benchmarks that assess RAG systems' capabilities under the multilingual multi-hop (MM-hop) QA setting. The second centers on the overreliance on LLMs' strong semantic understanding in English, which diminishes effectiveness in multilingual scenarios. To address these challenges, we first construct multilingual multi-hop QA benchmarks by translating English-only benchmarks into five languages, and then we propose DaPT, a novel multilingual RAG framework. DaPT generates sub-question graphs in parallel for both the source-language query and its English translation counterpart, then merges them before employing a bilingual retrieval-and-answer strategy to sequentially solve sub-questions. Our experimental results demonstrate that advanced RAG systems suffer from a significant performance imbalance in multilingual scenarios. Furthermore, our proposed method consistently yields more accurate and concise answers compared to the baselines, significantly enhancing RAG performance on this task. For instance, on the most challenging MuSiQue benchmark, DaPT achieves a relative improvement of 18.3\% in average EM score over the strongest baseline.

Yilin Wang, Heng Wang, Yuyang Bai et al.

In Large Language Models (LLMs) generation, there exist knowledge conflicts and scenarios where parametric knowledge contradicts knowledge provided in the context. Previous works studied tuning, decoding algorithms, or locating and editing context-aware neurons to adapt LLMs to be faithful to new contextual knowledge. However, they are usually inefficient or ineffective for large models, not workable for black-box models, or unable to continuously adjust LLMs' sensitivity to the knowledge provided in the context. To mitigate these problems, we propose CSKS (Continuously Steering Knowledge Sensitivity), a simple framework that can steer LLMs' sensitivity to contextual knowledge continuously at a lightweight cost. Specifically, we tune two small LMs (i.e. proxy models) and use the difference in their output distributions to shift the original distribution of an LLM without modifying the LLM weights. In the evaluation process, we not only design synthetic data and fine-grained metrics to measure models' sensitivity to contextual knowledge but also use a real conflict dataset to validate CSKS's practical efficacy. Extensive experiments demonstrate that our framework achieves continuous and precise control over LLMs' sensitivity to contextual knowledge, enabling both increased sensitivity and reduced sensitivity, thereby allowing LLMs to prioritize either contextual or parametric knowledge as needed flexibly. Our data and code are available at https://github.com/OliveJuiceLin/CSKS.

Thao Nguyen, Sicheng Mo, Krishna Kumar Singh et al.

Humans do not just see attribute similarity -- we also see relational similarity. An apple is like a peach because both are reddish fruit, but the Earth is also like a peach: its crust, mantle, and core correspond to the peach's skin, flesh, and pit. This ability to perceive and recognize relational similarity, is arguable by cognitive scientist to be what distinguishes humans from other species. Yet, all widely used visual similarity metrics today (e.g., LPIPS, CLIP, DINO) focus solely on perceptual attribute similarity and fail to capture the rich, often surprising relational similarities that humans perceive. How can we go beyond the visible content of an image to capture its relational properties? How can we bring images with the same relational logic closer together in representation space? To answer these questions, we first formulate relational image similarity as a measurable problem: two images are relationally similar when their internal relations or functions among visual elements correspond, even if their visual attributes differ. We then curate 114k image-caption dataset in which the captions are anonymized -- describing the underlying relational logic of the scene rather than its surface content. Using this dataset, we finetune a Vision-Language model to measure the relational similarity between images. This model serves as the first step toward connecting images by their underlying relational structure rather than their visible appearance. Our study shows that while relational similarity has a lot of real-world applications, existing image similarity models fail to capture it -- revealing a critical gap in visual computing.

Qi Zheng, Li-Heng Chen, Chenlong He et al.

Although there have been notable advancements in video compression technologies in recent years, banding artifacts remain a serious issue affecting the quality of compressed videos, particularly on smooth regions of high-definition videos. Noticeable banding artifacts can severely impact the perceptual quality of videos viewed on a high-end HDTV or high-resolution screen. Hence, there is a pressing need for a systematic investigation of the banding video quality assessment problem for advanced video codecs. Given that the existing publicly available datasets for studying banding artifacts are limited to still picture data only, which cannot account for temporal banding dynamics, we have created a first-of-a-kind open video dataset, dubbed LIVE-YT-Banding, which consists of 160 videos generated by four different compression parameters using the AV1 video codec. A total of 7,200 subjective opinions are collected from a cohort of 45 human subjects. To demonstrate the value of this new resources, we tested and compared a variety of models that detect banding occurrences, and measure their impact on perceived quality. Among these, we introduce an effective and efficient new no-reference (NR) video quality evaluator which we call CBAND. CBAND leverages the properties of the learned statistics of natural images expressed in the embeddings of deep neural networks. Our experimental results show that the perceptual banding prediction performance of CBAND significantly exceeds that of previous state-of-the-art models, and is also orders of magnitude faster. Moreover, CBAND can be employed as a differentiable loss function to optimize video debanding models. The LIVE-YT-Banding database, code, and pre-trained model are all publically available at https://github.com/uniqzheng/CBAND.

Shengtao Zhang, Haokai Zhang, Shiqi Lou et al.

Dynamic node classification is critical for modeling evolving systems like financial transactions and academic collaborations. In such systems, dynamically capturing node information changes is critical for dynamic node classification, which usually requires all labels at every timestamp. However, it is difficult to collect all dynamic labels in real-world scenarios due to high annotation costs and label uncertainty (e.g., ambiguous or delayed labels in fraud detection). In contrast, final timestamp labels are easier to obtain as they rely on complete temporal patterns and are usually maintained as a unique label for each user in many open platforms, without tracking the history data. To bridge this gap, we propose PTCL(Pseudo-label Temporal Curriculum Learning), a pioneering method addressing label-limited dynamic node classification where only final labels are available. PTCL introduces: (1) a temporal decoupling architecture separating the backbone (learning time-aware representations) and decoder (strictly aligned with final labels), which generate pseudo-labels, and (2) a Temporal Curriculum Learning strategy that prioritizes pseudo-labels closer to the final timestamp by assigning them higher weights using an exponentially decaying function. We contribute a new academic dataset (CoOAG), capturing long-range research interest in dynamic graph. Experiments across real-world scenarios demonstrate PTCL's consistent superiority over other methods adapted to this task. Beyond methodology, we propose a unified framework FLiD (Framework for Label-Limited Dynamic Node Classification), consisting of a complete preparation workflow, training pipeline, and evaluation standards, and supporting various models and datasets. The code can be found at https://github.com/3205914485/FLiD.

Pavan C. Madhusudana, Neil Birkbeck, Yilin Wang et al.

We consider the problem of obtaining image quality representations in a self-supervised manner. We use prediction of distortion type and degree as an auxiliary task to learn features from an unlabeled image dataset containing a mixture of synthetic and realistic distortions. We then train a deep Convolutional Neural Network (CNN) using a contrastive pairwise objective to solve the auxiliary problem. We refer to the proposed training framework and resulting deep IQA model as the CONTRastive Image QUality Evaluator (CONTRIQUE). During evaluation, the CNN weights are frozen and a linear regressor maps the learned representations to quality scores in a No-Reference (NR) setting. We show through extensive experiments that CONTRIQUE achieves competitive performance when compared to state-of-the-art NR image quality models, even without any additional fine-tuning of the CNN backbone. The learned representations are highly robust and generalize well across images afflicted by either synthetic or authentic distortions. Our results suggest that powerful quality representations with perceptual relevance can be obtained without requiring large labeled subjective image quality datasets. The implementations used in this paper are available at \url{https://github.com/pavancm/CONTRIQUE}.

Yifan Jiang, He Zhang, Jianming Zhang et al.

Image harmonization aims to improve the quality of image compositing by matching the "appearance" (\eg, color tone, brightness and contrast) between foreground and background images. However, collecting large-scale annotated datasets for this task requires complex professional retouching. Instead, we propose a novel Self-Supervised Harmonization framework (SSH) that can be trained using just "free" natural images without being edited. We reformulate the image harmonization problem from a representation fusion perspective, which separately processes the foreground and background examples, to address the background occlusion issue. This framework design allows for a dual data augmentation method, where diverse [foreground, background, pseudo GT] triplets can be generated by cropping an image with perturbations using 3D color lookup tables (LUTs). In addition, we build a real-world harmonization dataset as carefully created by expert users, for evaluation and benchmarking purposes. Our results show that the proposed self-supervised method outperforms previous state-of-the-art methods in terms of reference metrics, visual quality, and subject user study. Code and dataset are available at \url{https://github.com/VITA-Group/SSHarmonization}.

Zhengzhong Tu, Xiangxu Yu, Yilin Wang et al.

Blind or no-reference video quality assessment of user-generated content (UGC) has become a trending, challenging, heretofore unsolved problem. Accurate and efficient video quality predictors suitable for this content are thus in great demand to achieve more intelligent analysis and processing of UGC videos. Previous studies have shown that natural scene statistics and deep learning features are both sufficient to capture spatial distortions, which contribute to a significant aspect of UGC video quality issues. However, these models are either incapable or inefficient for predicting the quality of complex and diverse UGC videos in practical applications. Here we introduce an effective and efficient video quality model for UGC content, which we dub the Rapid and Accurate Video Quality Evaluator (RAPIQUE), which we show performs comparably to state-of-the-art (SOTA) models but with orders-of-magnitude faster runtime. RAPIQUE combines and leverages the advantages of both quality-aware scene statistics features and semantics-aware deep convolutional features, allowing us to design the first general and efficient spatial and temporal (space-time) bandpass statistics model for video quality modeling. Our experimental results on recent large-scale UGC video quality databases show that RAPIQUE delivers top performances on all the datasets at a considerably lower computational expense. We hope this work promotes and inspires further efforts towards practical modeling of video quality problems for potential real-time and low-latency applications. To promote public usage, an implementation of RAPIQUE has been made freely available online: \url{https://github.com/vztu/RAPIQUE}.

Qihang Yu, Jianming Zhang, He Zhang et al.

We propose Mask Guided (MG) Matting, a robust matting framework that takes a general coarse mask as guidance. MG Matting leverages a network (PRN) design which encourages the matting model to provide self-guidance to progressively refine the uncertain regions through the decoding process. A series of guidance mask perturbation operations are also introduced in the training to further enhance its robustness to external guidance. We show that PRN can generalize to unseen types of guidance masks such as trimap and low-quality alpha matte, making it suitable for various application pipelines. In addition, we revisit the foreground color prediction problem for matting and propose a surprisingly simple improvement to address the dataset issue. Evaluation on real and synthetic benchmarks shows that MG Matting achieves state-of-the-art performance using various types of guidance inputs. Code and models are available at https://github.com/yucornetto/MGMatting.

Pavan C. Madhusudana, Neil Birkbeck, Yilin Wang et al.

We consider the problem of conducting frame rate dependent video quality assessment (VQA) on videos of diverse frame rates, including high frame rate (HFR) videos. More generally, we study how perceptual quality is affected by frame rate, and how frame rate and compression combine to affect perceived quality. We devise an objective VQA model called Space-Time GeneRalized Entropic Difference (GREED) which analyzes the statistics of spatial and temporal band-pass video coefficients. A generalized Gaussian distribution (GGD) is used to model band-pass responses, while entropy variations between reference and distorted videos under the GGD model are used to capture video quality variations arising from frame rate changes. The entropic differences are calculated across multiple temporal and spatial subbands, and merged using a learned regressor. We show through extensive experiments that GREED achieves state-of-the-art performance on the LIVE-YT-HFR Database when compared with existing VQA models. The features used in GREED are highly generalizable and obtain competitive performance even on standard, non-HFR VQA databases. The implementation of GREED has been made available online: https://github.com/pavancm/GREED

Shikun Liu, Zhe Lin, Yilin Wang et al.

We present a novel resizing module for neural networks: shape adaptor, a drop-in enhancement built on top of traditional resizing layers, such as pooling, bilinear sampling, and strided convolution. Whilst traditional resizing layers have fixed and deterministic reshaping factors, our module allows for a learnable reshaping factor. Our implementation enables shape adaptors to be trained end-to-end without any additional supervision, through which network architectures can be optimised for each individual task, in a fully automated way. We performed experiments across seven image classification datasets, and results show that by simply using a set of our shape adaptors instead of the original resizing layers, performance increases consistently over human-designed networks, across all datasets. Additionally, we show the effectiveness of shape adaptors on two other applications: network compression and transfer learning. The source code is available at: https://github.com/lorenmt/shape-adaptor.

Zhengzhong Tu, Yilin Wang, Neil Birkbeck et al.

Recent years have witnessed an explosion of user-generated content (UGC) videos shared and streamed over the Internet, thanks to the evolution of affordable and reliable consumer capture devices, and the tremendous popularity of social media platforms. Accordingly, there is a great need for accurate video quality assessment (VQA) models for UGC/consumer videos to monitor, control, and optimize this vast content. Blind quality prediction of in-the-wild videos is quite challenging, since the quality degradations of UGC content are unpredictable, complicated, and often commingled. Here we contribute to advancing the UGC-VQA problem by conducting a comprehensive evaluation of leading no-reference/blind VQA (BVQA) features and models on a fixed evaluation architecture, yielding new empirical insights on both subjective video quality studies and VQA model design. By employing a feature selection strategy on top of leading VQA model features, we are able to extract 60 of the 763 statistical features used by the leading models to create a new fusion-based BVQA model, which we dub the \textbf{VID}eo quality \textbf{EVAL}uator (VIDEVAL), that effectively balances the trade-off between VQA performance and efficiency. Our experimental results show that VIDEVAL achieves state-of-the-art performance at considerably lower computational cost than other leading models. Our study protocol also defines a reliable benchmark for the UGC-VQA problem, which we believe will facilitate further research on deep learning-based VQA modeling, as well as perceptually-optimized efficient UGC video processing, transcoding, and streaming. To promote reproducible research and public evaluation, an implementation of VIDEVAL has been made available online: \url{https://github.com/tu184044109/VIDEVAL_release}.

Jian Zou, Xiaoyu Xu, Zhihua Wang et al.

Learning-based video quality assessment (VQA) has advanced rapidly, yet progress is increasingly constrained by a disconnect between model design and dataset curation. Model-centric approaches often iterate on fixed benchmarks, while data-centric efforts collect new human labels without systematically targeting the weaknesses of existing VQA models. Here, we describe MDS-VQA, a model-informed data selection mechanism for curating unlabeled videos that are both difficult for the base VQA model and diverse in content. Difficulty is estimated by a failure predictor trained with a ranking objective, and diversity is measured using deep semantic video features, with a greedy procedure balancing the two under a constrained labeling budget. Experiments across multiple VQA datasets and models demonstrate that MDS-VQA identifies diverse, challenging samples that are particularly informative for active fine-tuning. With only a 5% selected subset per target domain, the fine-tuned model improves mean SRCC from 0.651 to 0.722 and achieves the top gMAD rank, indicating strong adaptation and generalization.

Xi Chen, Zhifei Zhang, He Zhang et al.

We introduce UniReal, a unified framework designed to address various image generation and editing tasks. Existing solutions often vary by tasks, yet share fundamental principles: preserving consistency between inputs and outputs while capturing visual variations. Inspired by recent video generation models that effectively balance consistency and variation across frames, we propose a unifying approach that treats image-level tasks as discontinuous video generation. Specifically, we treat varying numbers of input and output images as frames, enabling seamless support for tasks such as image generation, editing, customization, composition, etc. Although designed for image-level tasks, we leverage videos as a scalable source for universal supervision. UniReal learns world dynamics from large-scale videos, demonstrating advanced capability in handling shadows, reflections, pose variation, and object interaction, while also exhibiting emergent capability for novel applications.

Zirui Wang, Zhizhou Sha, Zheng Ding et al. · princeton

We present TokenCompose, a Latent Diffusion Model for text-to-image generation that achieves enhanced consistency between user-specified text prompts and model-generated images. Despite its tremendous success, the standard denoising process in the Latent Diffusion Model takes text prompts as conditions only, absent explicit constraint for the consistency between the text prompts and the image contents, leading to unsatisfactory results for composing multiple object categories. TokenCompose aims to improve multi-category instance composition by introducing the token-wise consistency terms between the image content and object segmentation maps in the finetuning stage. TokenCompose can be applied directly to the existing training pipeline of text-conditioned diffusion models without extra human labeling information. By finetuning Stable Diffusion, the model exhibits significant improvements in multi-category instance composition and enhanced photorealism for its generated images. Project link: https://mlpc-ucsd.github.io/TokenCompose

Haiyang Xu, Yu Lei, Zeyuan Chen et al.

We present Bayesian Diffusion Models (BDM), a prediction algorithm that performs effective Bayesian inference by tightly coupling the top-down (prior) information with the bottom-up (data-driven) procedure via joint diffusion processes. We show the effectiveness of BDM on the 3D shape reconstruction task. Compared to prototypical deep learning data-driven approaches trained on paired (supervised) data-labels (e.g. image-point clouds) datasets, our BDM brings in rich prior information from standalone labels (e.g. point clouds) to improve the bottom-up 3D reconstruction. As opposed to the standard Bayesian frameworks where explicit prior and likelihood are required for the inference, BDM performs seamless information fusion via coupled diffusion processes with learned gradient computation networks. The specialty of our BDM lies in its capability to engage the active and effective information exchange and fusion of the top-down and bottom-up processes where each itself is a diffusion process. We demonstrate state-of-the-art results on both synthetic and real-world benchmarks for 3D shape reconstruction.

Yilin Wang, Haojie Huang, Chen Li et al.

Sand painting is a process-driven art where visual appearance emerges from granular accumulation. Given a single image, reconstructing a plausible sand painting process requires modeling coherent stroke structures and material-dependent effects. Existing methods, including stroke-based optimization and diffusion-based video synthesis, often lack structural coherence and material consistency, leading to unrealistic drawing sequences. We present SandSim, a framework that reconstructs a sand painting process from a single image. We introduce a curve-guided Gaussian representation that models strokes as sequences of anisotropic primitives along continuous trajectories, whose smooth kernels capture the soft boundaries of sand strokes and enable coherent stroke formation. We further adopt a subtractive compositing scheme to model light attenuation during sand accumulation. We incorporate a semantic-guided planning module for scene decomposition and drawing order inference. Our framework jointly optimizes stroke geometry and appearance and can be integrated with a physics-based simulator for interactive sand dynamics and editing. Experiments show that our method produces temporally coherent and visually realistic results, achieving improved reconstruction quality and perceptual fidelity compared to existing approaches.

Yuchun Fan, Yongyu Mu, Yilin Wang et al.

Despite the significant improvements achieved by large language models (LLMs) in English reasoning tasks, these models continue to struggle with multilingual reasoning. Recent studies leverage a full-parameter and two-stage training paradigm to teach models to first understand non-English questions and then reason. However, this method suffers from both substantial computational resource computing and catastrophic forgetting. The fundamental cause is that, with the primary goal of enhancing multilingual comprehension, an excessive number of irrelevant layers and parameters are tuned during the first stage. Given our findings that the representation learning of languages is merely conducted in lower-level layers, we propose an efficient multilingual reasoning alignment approach that precisely identifies and fine-tunes the layers responsible for handling multilingualism. Experimental results show that our method, SLAM, only tunes 6 layers' feed-forward sub-layers including 6.5-8% of all parameters within 7B and 13B LLMs, achieving superior average performance than all strong baselines across 10 languages. Meanwhile, SLAM only involves one training stage, reducing training time by 4.1-11.9 compared to the two-stage method.

Hang Hua, Qing Liu, Lingzhi Zhang et al.

The advent of large Vision-Language Models (VLMs) has significantly advanced multimodal tasks, enabling more sophisticated and accurate reasoning across various applications, including image and video captioning, visual question answering, and cross-modal retrieval. Despite their superior capabilities, VLMs struggle with fine-grained image regional composition information perception. Specifically, they have difficulty accurately aligning the segmentation masks with the corresponding semantics and precisely describing the compositional aspects of the referred regions. However, compositionality - the ability to understand and generate novel combinations of known visual and textual components - is critical for facilitating coherent reasoning and understanding across modalities by VLMs. To address this issue, we propose FINECAPTION, a novel VLM that can recognize arbitrary masks as referential inputs and process high-resolution images for compositional image captioning at different granularity levels. To support this endeavor, we introduce COMPOSITIONCAP, a new dataset for multi-grained region compositional image captioning, which introduces the task of compositional attribute-aware regional image captioning. Empirical results demonstrate the effectiveness of our proposed model compared to other state-of-the-art VLMs. Additionally, we analyze the capabilities of current VLMs in recognizing various visual prompts for compositional region image captioning, highlighting areas for improvement in VLM design and training.

Jing Gu, Nanxuan Zhao, Wei Xiong et al.

Effective editing of personal content holds a pivotal role in enabling individuals to express their creativity, weaving captivating narratives within their visual stories, and elevate the overall quality and impact of their visual content. Therefore, in this work, we introduce SwapAnything, a novel framework that can swap any objects in an image with personalized concepts given by the reference, while keeping the context unchanged. Compared with existing methods for personalized subject swapping, SwapAnything has three unique advantages: (1) precise control of arbitrary objects and parts rather than the main subject, (2) more faithful preservation of context pixels, (3) better adaptation of the personalized concept to the image. First, we propose targeted variable swapping to apply region control over latent feature maps and swap masked variables for faithful context preservation and initial semantic concept swapping. Then, we introduce appearance adaptation, to seamlessly adapt the semantic concept into the original image in terms of target location, shape, style, and content during the image generation process. Extensive results on both human and automatic evaluation demonstrate significant improvements of our approach over baseline methods on personalized swapping. Furthermore, SwapAnything shows its precise and faithful swapping abilities across single object, multiple objects, partial object, and cross-domain swapping tasks. SwapAnything also achieves great performance on text-based swapping and tasks beyond swapping such as object insertion.

Yilin Wang, Minghao Hu, Zhen Huang et al.

The goal of knowledge graph completion (KGC) is to predict missing facts among entities. Previous methods for KGC re-ranking are mostly built on non-generative language models to obtain the probability of each candidate. Recently, generative large language models (LLMs) have shown outstanding performance on several tasks such as information extraction and dialog systems. Leveraging them for KGC re-ranking is beneficial for leveraging the extensive pre-trained knowledge and powerful generative capabilities. However, it may encounter new problems when accomplishing the task, namely mismatch, misordering and omission. To this end, we introduce KC-GenRe, a knowledge-constrained generative re-ranking method based on LLMs for KGC. To overcome the mismatch issue, we formulate the KGC re-ranking task as a candidate identifier sorting generation problem implemented by generative LLMs. To tackle the misordering issue, we develop a knowledge-guided interactive training method that enhances the identification and ranking of candidates. To address the omission issue, we design a knowledge-augmented constrained inference method that enables contextual prompting and controlled generation, so as to obtain valid rankings. Experimental results show that KG-GenRe achieves state-of-the-art performance on four datasets, with gains of up to 6.7% and 7.7% in the MRR and Hits@1 metric compared to previous methods, and 9.0% and 11.1% compared to that without re-ranking. Extensive analysis demonstrates the effectiveness of components in KG-GenRe.

Yilin Wang, Peixuan Lei, Jie Song et al.

Time-series data are critical in diverse applications, such as industrial monitoring, medical diagnostics, and climate research. However, effectively integrating these high-dimensional temporal signals with natural language for dynamic, interactive tasks remains a significant challenge. To address this, we introduce the Time-Series Question Answering (Time-Series QA) task and release EngineMT-QA, the first large-scale, multi-task, temporal-textual QA dataset designed to capture complex interactions between time-series signals and natural language. Building on this resource, we propose the Instruct Time Transformer (ITFormer), a novel framework that bridges time-series encoders with frozen large language models (LLMs). ITFormer effectively extracts, aligns, and fuses temporal and textual features, achieving a strong improvement in QA accuracy over strong baselines with fewer than 1\% additional trainable parameters. By combining computational efficiency with robust cross-modal modeling, our work establishes a adaptable paradigm for integrating temporal data with natural language, paving the way for new research and applications in multi-modal AI. More details about the project, including datasets and code, are available at: https://pandalin98.github.io/itformer_site/

Wen Wen, Yilin Wang, Neil Birkbeck et al.

The rise of short-form videos, characterized by diverse content, editing styles, and artifacts, poses substantial challenges for learning-based blind video quality assessment (BVQA) models. Multimodal large language models (MLLMs), renowned for their superior generalization capabilities, present a promising solution. This paper focuses on effectively leveraging a pretrained MLLM for short-form video quality assessment, regarding the impacts of pre-processing and response variability, and insights on combining the MLLM with BVQA models. We first investigated how frame pre-processing and sampling techniques influence the MLLM's performance. Then, we introduced a lightweight learning-based ensemble method that adaptively integrates predictions from the MLLM and state-of-the-art BVQA models. Our results demonstrated superior generalization performance with the proposed ensemble approach. Furthermore, the analysis of content-aware ensemble weights highlighted that some video characteristics are not fully represented by existing BVQA models, revealing potential directions to improve BVQA models further.

Ye Wang, Ruiqi Liu, Jiang Lin et al.

In this paper, we introduce OmniStyle-1M, a large-scale paired style transfer dataset comprising over one million content-style-stylized image triplets across 1,000 diverse style categories, each enhanced with textual descriptions and instruction prompts. We show that OmniStyle-1M can not only enable efficient and scalable of style transfer models through supervised training but also facilitate precise control over target stylization. Especially, to ensure the quality of the dataset, we introduce OmniFilter, a comprehensive style transfer quality assessment framework, which filters high-quality triplets based on content preservation, style consistency, and aesthetic appeal. Building upon this foundation, we propose OmniStyle, a framework based on the Diffusion Transformer (DiT) architecture designed for high-quality and efficient style transfer. This framework supports both instruction-guided and image-guided style transfer, generating high resolution outputs with exceptional detail. Extensive qualitative and quantitative evaluations demonstrate OmniStyle's superior performance compared to existing approaches, highlighting its efficiency and versatility. OmniStyle-1M and its accompanying methodologies provide a significant contribution to advancing high-quality style transfer, offering a valuable resource for the research community.