Peer Revenue Past citations are no guarantee of future returns

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

Computer Science Computer Vision

cs.CV Watch Computer Vision (sign in to use)

Thursday 4 December

Computer Vision

SP-Det: Self-Prompted Dual-Text Fusion for Generalized Multi-Label Lesion Detection

Automated lesion detection in chest X-rays has demonstrated significant potential for improving clinical diagnosis by precisely localizing pathological abnormalities. While recent promptable detection frameworks have achieved remarkable accuracy in target localization, existing methods typically rely on manual annotations as prompts, which are labor-intensive and impractical for clinical applications. To address this limitation, we propose SP-Det, a novel self-prompted detection framework that automatically generates rich textual context to guide multi-label lesion detection without requiring expert annotations. Specifically, we introduce an expert-free dual-text prompt generator (DTPG) that leverages two complementary textual modalities: semantic context prompts that capture global pathological patterns and disease beacon prompts that focus on disease-specific manifestations. Moreover, we devise a bidirectional feature enhancer (BFE) that synergistically integrates comprehensive diagnostic context with disease-specific embeddings to significantly improve feature representation and detection accuracy. Extensive experiments on two chest X-ray datasets with diverse thoracic disease categories demonstrate that our SP-Det framework outperforms state-of-the-art detection methods while completely eliminating the dependency on expert-annotated prompts compared to existing promptable architectures.
Read the full abstract

All authors Qing Xu, Yanqian Wang, Xiangjian Hea, Yue Li, Yixuan Zhang, Rong Qu, Wenting Duan, Zhen Chen

Computer Vision

Contact-Aware Refinement of Human Pose Pseudo-Ground Truth via Bioimpedance Sensing

Capturing accurate 3D human pose in the wild would provide valuable data for training pose estimation and motion generation methods. While video-based estimation approaches have become increasingly accurate, they often fail in common scenarios involving self-contact, such as a hand touching the face. In contrast, wearable bioimpedance sensing can cheaply and unobtrusively measure ground-truth skin-to-skin contact. Consequently, we propose a novel framework that combines visual pose estimators with bioimpedance sensing to capture the 3D pose of people by taking self-contact into account. Our method, BioTUCH, initializes the pose using an off-the-shelf estimator and introduces contact-aware pose optimization during measured self-contact: reprojection error and deviations from the input estimate are minimized while enforcing vertex proximity constraints. We validate our approach using a new dataset of synchronized RGB video, bioimpedance measurements, and 3D motion capture. Testing with three input pose estimators, we demonstrate an average of 11.7% improvement in reconstruction accuracy. We also present a miniature wearable bioimpedance sensor that enables efficient large-scale collection of contact-aware training data for improving pose estimation and generation using BioTUCH. Code and data are available at biotuch.is.tue.mpg.de
Read the full abstract

All authors Maria-Paola Forte, Nikos Athanasiou, Giulia Ballardini, Jan Ulrich Bartels, Katherine J. Kuchenbecker, Michael J. Black

Computer Vision

Autoregressive Image Generation Needs Only a Few Lines of Cached Tokens

Autoregressive (AR) visual generation has emerged as a powerful paradigm for image and multimodal synthesis, owing to its scalability and generality. However, existing AR image generation suffers from severe memory bottlenecks due to the need to cache all previously generated visual tokens during decoding, leading to both high storage requirements and low throughput. In this paper, we introduce LineAR, a novel, training-free progressive key-value (KV) cache compression pipeline for autoregressive image generation. By fully exploiting the intrinsic characteristics of visual attention, LineAR manages the cache at the line level using a 2D view, preserving the visual dependency regions while progressively evicting less-informative tokens that are harmless for subsequent line generation, guided by inter-line attention. LineAR enables efficient autoregressive (AR) image generation by utilizing only a few lines of cache, achieving both memory savings and throughput speedup, while maintaining or even improving generation quality. Extensive experiments across six autoregressive image generation models, including class-conditional and text-to-image generation, validate its effectiveness and generality. LineAR improves ImageNet FID from 2.77 to 2.68 and COCO FID from 23.85 to 22.86 on LlamaGen-XL and Janus-Pro-1B, while retaining only 1/6 KV cache. It also improves DPG on Lumina-mGPT-768 with just 1/8 KV cache. Additionally, LineAR achieves significant memory and throughput gains, including up to 67.61% memory reduction and 7.57x speedup on LlamaGen-XL, and 39.66% memory reduction and 5.62x speedup on Janus-Pro-7B.
Read the full abstract

All authors Ziran Qin, Youru Lv, Mingbao Lin, Zeren Zhang, Chanfan Gan, Tieyuan Chen, Weiyao Lin

Computer Vision

A Sanity Check for Multi-In-Domain Face Forgery Detection in the Real World

Existing methods for deepfake detection aim to develop generalizable detectors. Although "generalizable" is the ultimate target once and for all, with limited training forgeries and domains, it appears idealistic to expect generalization that covers entirely unseen variations, especially given the diversity of real-world deepfakes. Therefore, introducing large-scale multi-domain data for training can be feasible and important for real-world applications. However, within such a multi-domain scenario, the differences between multiple domains, rather than the subtle real/fake distinctions, dominate the feature space. As a result, despite detectors being able to relatively separate real and fake within each domain (i.e., high AUC), they struggle with single-image real/fake judgments in domain-unspecified conditions (i.e., low ACC). In this paper, we first define a new research paradigm named Multi-In-Domain Face Forgery Detection (MID-FFD), which includes sufficient volumes of real-fake domains for training. Then, the detector should provide definitive real-fake judgments to the domain-unspecified inputs, which simulate the frame-by-frame independent detection scenario in the real world. Meanwhile, to address the domain-dominant issue, we propose a model-agnostic framework termed DevDet (Developer for Detector) to amplify real/fake differences and make them dominant in the feature space. DevDet consists of a Face Forgery Developer (FFDev) and a Dose-Adaptive detector Fine-Tuning strategy (DAFT). Experiments demonstrate our superiority in predicting real-fake under the MID-FFD scenario while maintaining original generalization ability to unseen data.
Read the full abstract

All authors Jikang Cheng, Renye Yan, Zhiyuan Yan, Yaozhong Gan, Xueyi Zhang, Zhongyuan Wang, Wei Peng, Ling Liang

Computer Vision

BIM-Native Tokenization for Constraint-Aware Room Layout Synthesis

We present a BIM-native tokenization for room-level layout synthesis in Building Information Modeling (BIM) scenes. The core contribution is representational: we encode each room as a sequence of BIM-Token Bundles, realized as columns of a sparse attribute-feature matrix that unifies categorical and continuous attributes of walls, openings, and entities under wall-referenced (translation/scale-invariant) coordinates. A mixed-type embedding module produces a unified token vector from this matrix; a single Transformer backbone is then trained in two modes: encoder-only for room embeddings and retrieval, and encoder-decoder for autoregressive entity placement, which we call Data-Driven Entity Prediction (DDEP). On a controlled same-data benchmark with shared ontology and evaluation harness, DDEP outperforms ATISS and BLT baselines bridged into our representation, with ablations identifying joint continuous-feature embedding and entity ordering as primary drivers. Encoder embeddings cluster rooms by type more tightly than large general-purpose text encoders, which in turn retain an edge on within-type ranking. We frame this work as evidence that modestly sized, domain-specific sequence models over well-designed BIM tokenizations are a useful primitive for constraint-aware spatial generation, complementary to general-purpose LLMs/VLMs which we also benchmark.
Read the full abstract

All authors Manuel Ladron de Guevara, Jinmo Rhee, Ardavan Bidgoli, Vaidas Razgaitis, Michael Bergin

Computer Vision

FreeGen: Feed-Forward Reconstruction-Generation Co-Training for Free-Viewpoint Driving Scene Synthesis

Closed-loop simulation and scalable pre-training for autonomous driving require synthesizing free-viewpoint driving scenes. However, existing datasets and generative pipelines rarely provide consistent off-trajectory observations, limiting large-scale evaluation and training. While recent generative models demonstrate strong visual realism, they struggle to jointly achieve interpolation consistency and extrapolation realism without per-scene optimization. To address this, we propose FreeGen, a feed-forward reconstruction-generation co-training framework for free-viewpoint driving scene synthesis. The reconstruction model provides stable geometric representations to ensure interpolation consistency, while the generation model performs geometry-aware enhancement to improve realism at unseen viewpoints. Through co-training, generative priors are distilled into the reconstruction model to improve off-trajectory rendering, and the refined geometry in turn offers stronger structural guidance for generation. Experiments demonstrate that FreeGen achieves state-of-the-art performance for free-viewpoint driving scene synthesis.
Read the full abstract

Computer Vision

LatentFM: A Latent Flow Matching Approach for Generative Medical Image Segmentation

Generative models have achieved remarkable progress with the emergence of flow matching (FM). It has demonstrated strong generative capabilities and attracted significant attention as a simulation-free flow-based framework capable of learning exact data densities. Motivated by these advances, we propose LatentFM, a flow-based model operating in the latent space for medical image segmentation. To model the data distribution, we first design two variational autoencoders (VAEs) to encode both medical images and their corresponding masks into a lower-dimensional latent space. We then estimate a conditional velocity field that guides the flow based on the input image. By sampling multiple latent representations, our method synthesizes diverse segmentation outputs whose pixel-wise variance reliably captures the underlying data distribution, enabling both highly accurate and uncertainty-aware predictions. Furthermore, we generate confidence maps that quantify the model certainty, providing clinicians with richer information for deeper analysis. We conduct experiments on two datasets, ISIC-2018 and CVC-Clinic, and compare our method with several prior baselines, including both deterministic and generative approach models. Through comprehensive evaluations, both qualitative and quantitative results show that our approach achieves superior segmentation accuracy while remaining highly efficient in the latent space.
Read the full abstract

All authors Ngoc Huynh Trinh, Hoang Anh Nguyen Kim, Hai Toan Nguyen, Quoc Long Tran

Computer Vision

RobustSplat++: Decoupling Densification, Dynamics, and Illumination for In-the-Wild 3DGS

3D Gaussian Splatting (3DGS) has gained significant attention for its real-time, photo-realistic rendering in novel-view synthesis and 3D modeling. However, existing methods struggle with accurately modeling in-the-wild scenes affected by transient objects and illuminations, leading to artifacts in the rendered images. We identify that the Gaussian densification process, while enhancing scene detail capture, unintentionally contributes to these artifacts by growing additional Gaussians that model transient disturbances and illumination variations. To address this, we propose RobustSplat++, a robust solution based on several critical designs. First, we introduce a delayed Gaussian growth strategy that prioritizes optimizing static scene structure before allowing Gaussian splitting/cloning, mitigating overfitting to transient objects in early optimization. Second, we design a scale-cascaded mask bootstrapping approach that first leverages lower-resolution feature similarity supervision for reliable initial transient mask estimation, taking advantage of its stronger semantic consistency and robustness to noise, and then progresses to high-resolution supervision to achieve more precise mask prediction. Third, we incorporate the delayed Gaussian growth strategy and mask bootstrapping with appearance modeling to handling in-the-wild scenes including transients and illuminations. Extensive experiments on multiple challenging datasets show that our method outperforms existing methods, clearly demonstrating the robustness and effectiveness of our method.
Read the full abstract

All authors Chuanyu Fu, Guanying Chen, Yuqi Zhang, Kunbin Yao, Yuan Xiong, Chuan Huang, Shuguang Cui, Yasuyuki Matsushita, Xiaochun Cao

Computer Vision · Cross-listed from cs.SD

Shared Multi-modal Embedding Space for Face-Voice Association

The FAME 2026 challenge comprises two demanding tasks: training face-voice associations combined with a multilingual setting that includes testing on languages on which the model was not trained. Our approach consists of separate uni-modal processing pipelines with general face and voice feature extraction, complemented by additional age-gender feature extraction to support prediction. The resulting single-modal features are projected into a shared embedding space and trained with an Adaptive Angular Margin (AAM) loss. Our approach achieved first place in the FAME 2026 challenge, with an average Equal-Error Rate (EER) of 23.99%.
Read the full abstract

Computer Vision

EMMA: Efficient Multimodal Understanding, Generation, and Editing with a Unified Architecture

We propose EMMA, an efficient and unified architecture for multimodal understanding, generation and editing. Specifically, EMMA primarily consists of 1) An efficient autoencoder with a 32x compression ratio, which significantly reduces the number of tokens required for generation. This also ensures the training balance between understanding and generation tasks by applying the same compression ratio to images. 2) Channel-wise concatenation instead of token-wise concatenation among visual understanding and generation tokens, which further reduces the visual tokens in unified architectures. 3) A shared-and-decoupled network that enables mutual improvements across tasks while meeting the task-specific modeling requirements. 4) A mixture-of-experts mechanism adopted for visual understanding encoder, which substantially improves perceptual capabilities with a few parameters increase. Extensive experiments have shown that EMMA-4B can significantly outperform state-of-the-art unified multimodal approaches (e.g., BAGEL-7B) in both efficiency and performance, while also achieving competitive results compared to recent multimodal understanding and generation experts (e.g., Qwen3-VL and Qwen-Image). We believe that EMMA lays a solid foundation for the future development of unified multimodal architectures.
Read the full abstract

All authors Xin He, Longhui Wei, Jianbo Ouyang, Minghui Liao, Lingxi Xie, Qi Tian

Computer Vision

LaFiTe: A Generative Latent Field for 3D Native Texturing

Generating high-fidelity, seamless textures directly on 3D surfaces, what we term 3D-native texturing, remains a fundamental open challenge, with the potential to overcome long-standing limitations of UV-based and multi-view projection methods. However, existing native approaches are constrained by the absence of a powerful and versatile latent representation, which severely limits the fidelity and generality of their generated textures. We identify this representation gap as the principal barrier to further progress. We introduce LaFiTe, a framework that addresses this challenge by learning to generate textures as a 3D generative sparse latent color field. At its core, LaFiTe employs a variational autoencoder (VAE) to encode complex surface appearance into a sparse, structured latent space, which is subsequently decoded into a continuous color field. This representation achieves unprecedented fidelity, exceeding state-of-the-art methods by >10 dB PSNR in reconstruction, by effectively disentangling texture appearance from mesh topology and UV parameterization. Building upon this strong representation, a conditional rectified-flow model synthesizes high-quality, coherent textures across diverse styles and geometries. Extensive experiments demonstrate that LaFiTe not only sets a new benchmark for 3D-native texturing but also enables flexible downstream applications such as material synthesis and texture super-resolution, paving the way for the next generation of 3D content creation workflows.
Read the full abstract

All authors Chia-Hao Chen, Zi-Xin Zou, Yan-Pei Cao, Ze Yuan, Guan Luo, Xiaojuan Qi, Ding Liang, Song-Hai Zhang, Yuan-Chen Guo

Computer Vision · Cross-listed from cs.LG

MemLoRA: Distilling Expert Adapters for On-Device Memory Systems

Memory-augmented Large Language Models (LLMs) have demonstrated remarkable consistency during prolonged dialogues by storing relevant memories and incorporating them as context. Such memory-based personalization is also key in on-device settings that allow users to keep their conversations and data private. However, memory-augmented systems typically rely on LLMs that are too costly for local on-device deployment. Even though Small Language Models (SLMs) are more suitable for on-device inference than LLMs, they cannot achieve sufficient performance. Additionally, these LLM-based systems lack native visual capabilities, limiting their applicability in multimodal contexts. In this paper, we introduce (i) MemLoRA, a novel memory system that enables local deployment by equipping SLMs with specialized memory adapters, and (ii) its vision extension MemLoRA-V, which integrates small Vision-Language Models (SVLMs) to memory systems, enabling native visual understanding. Following knowledge distillation principles, each adapter is trained separately for specific memory operations$\unicode{x2013}$knowledge extraction, memory update, and memory-augmented generation. Equipped with memory adapters, small models enable accurate on-device memory operations without cloud dependency. On text-only operations, MemLoRA outperforms 10$\times$ larger baseline models (e.g., Gemma2-27B) and achieves performance comparable to 60$\times$ larger models (e.g., GPT-OSS-120B) on the LoCoMo benchmark. To evaluate visual understanding operations instead, we extend LoCoMo with challenging Visual Question Answering tasks that require direct visual reasoning. On this, our VLM-integrated MemLoRA-V shows massive improvements over caption-based approaches (81.3 vs. 23.7 accuracy) while keeping strong performance in text-based tasks, demonstrating the efficacy of our method in multimodal contexts.
Read the full abstract

All authors Massimo Bini, Ondrej Bohdal, Umberto Michieli, Zeynep Akata, Mete Ozay, Taha Ceritli

Computer Vision

Order Matters: 3D Shape Generation from Sequential VR Sketches

VR sketching lets users explore and iterate on ideas directly in 3D, offering a faster and more intuitive alternative to conventional CAD tools. However, existing sketch-to-shape models ignore the temporal ordering of strokes, discarding crucial cues about structure and design intent. We introduce VRSketch2Shape, the first framework and multi-category dataset for generating 3D shapes from sequential VR sketches. Our contributions are threefold: (i) an automated pipeline that generates sequential VR sketches from arbitrary shapes, (ii) a dataset of over 20k synthetic and 900 hand-drawn sketch-shape pairs across four categories, and (iii) an order-aware sketch encoder coupled with a diffusion-based 3D generator. Our approach yields higher geometric fidelity than prior work, generalizes effectively from synthetic to real sketches with minimal supervision, and performs well even on partial sketches. All data and models will be released open-source at https://chenyizi086.github.io/VRSketch2Shape_website.
Read the full abstract

All authors Yizi Chen, Sidi Wu, Tianyi Xiao, Nina Wiedemann, Loic Landrieu

Computer Vision

MT-Depth: Multi-task Instance feature analysis for the Depth Completion

Depth completion plays a vital role in 3D perception systems, especially in scenarios where sparse depth data must be densified for tasks such as autonomous driving, robotics, and augmented reality. While many existing approaches rely on semantic segmentation to guide depth completion, they often overlook the benefits of object-level understanding. In this work, we introduce an instance-aware depth completion framework that explicitly integrates binary instance masks as spatial priors to refine depth predictions. Our model combines four main components: a frozen YOLO V11 instance segmentation branch, a U-Net-based depth completion backbone, a cross-attention fusion module, and an attention-guided prediction head. The instance segmentation branch generates per-image foreground masks that guide the depth branch via cross-attention, allowing the network to focus on object-centric regions during refinement. We validate our method on the Virtual KITTI 2 dataset, showing that it achieves lower Root Mean Squared Error (RMSE) compared to both a U-Net-only baseline and previous semantic-guided methods, while maintaining competitive Mean Absolute Error (MAE). Qualitative and quantitative results demonstrate that the proposed model effectively enhances depth accuracy near object boundaries, occlusions, and thin structures. Our findings suggest that incorporating instance-aware cues offers a promising direction for improving depth completion without relying on dense semantic labels.
Read the full abstract

Computer Vision

E3AD: An Emotion-Aware Vision-Language-Action Model for Human-Centric End-to-End Autonomous Driving

End-to-end autonomous driving (AD) systems increasingly adopt vision-language-action (VLA) models, yet they typically ignore the passenger's emotional state, which is central to comfort and AD acceptance. We introduce Open-Domain End-to-End (OD-E2E) autonomous driving, where an autonomous vehicle (AV) must interpret free-form natural-language commands, infer the emotion, and plan a physically feasible trajectory. We propose E3AD, an emotion-aware VLA framework that augments semantic understanding with two cognitively inspired components: a continuous Valenc-Arousal-Dominance (VAD) emotion model that captures tone and urgency from language, and a dual-pathway spatial reasoning module that fuses egocentric and allocentric views for human-like spatial cognition. A consistency-oriented training scheme, combining modality pretraining with preference-based alignment, further enforces coherence between emotional intent and driving actions. Across real-world datasets, E3AD improves visual grounding and waypoint planning and achieves state-of-the-art (SOTA) VAD correlation for emotion estimation. These evaluation results show that injecting emotion into VLA-style driving yields more human-aligned grounding, planning, and feedback.
Read the full abstract

All authors Yihong Tang, Haicheng Liao, Tong Nie, Junlin He, Ao Qu, Kehua Chen, Wei Ma, Zhenning Li, Lijun Sun, Chengzhong Xu

Computer Vision

Measuring the Unspoken: A Disentanglement Model and Benchmark for Psychological Analysis in the Wild

Generative psychological analysis of in-the-wild conversations faces two fundamental challenges: (1) existing Vision-Language Models (VLMs) fail to resolve Articulatory-Affective Ambiguity, where visual patterns of speech mimic emotional expressions; and (2) progress is stifled by a lack of verifiable evaluation metrics capable of assessing visual grounding and reasoning depth. We propose a complete ecosystem to address these twin challenges. First, we introduce Multilevel Insight Network for Disentanglement(MIND), a novel hierarchical visual encoder that introduces a Status Judgment module to algorithmically suppress ambiguous lip features based on their temporal feature variance, achieving explicit visual disentanglement. Second, we construct ConvoInsight-DB, a new large-scale dataset with expert annotations for micro-expressions and deep psychological inference. Third, Third, we designed the Mental Reasoning Insight Rating Metric (PRISM), an automated dimensional framework that uses expert-guided LLM to measure the multidimensional performance of large mental vision models. On our PRISM benchmark, MIND significantly outperforms all baselines, achieving a +86.95% gain in micro-expression detection over prior SOTA. Ablation studies confirm that our Status Judgment disentanglement module is the most critical component for this performance leap. Our code has been opened.
Read the full abstract

All authors Yigui Feng, Qinglin Wang, Haotian Mo, Yang Liu, Ke Liu, Gencheng Liu, Xinhai Chen, Siqi Shen, Songzhu Mei, Jie Liu

Computer Vision · Cross-listed from eess.SP

CIG-MAE: Cross-Modal Information-Guided Masked Autoencoder for Self-Supervised WiFi Sensing

Human Action Recognition using WiFi Channel State Information (CSI) has emerged as an attractive alternative to vision-based methods due to its ubiquity, device-agnostic nature, and inherent privacy-preserving capabilities. However, the high cost of manual annotation and the limited scale of publicly available CSI datasets restrict the performance of supervised approaches. Self-supervised learning (SSL) offers a promising avenue, but existing contrastive paradigms rely on data augmentations that conflict with the physical semantics of radio signals and require large-batch training, making them poorly suited for CSI. To overcome these challenges, we introduce CIG-MAE – a Cross-modal Information-Guided Masked Autoencoder – that reconstructs both the amplitude and phase of CSI using a symmetric dual-stream architecture with a high masking ratio. Specifically, we propose an Adaptive Information-Guided Masking strategy that dynamically allocates attention to time-frequency regions with high information density to improve learning efficiency, and incorporate a Barlow Twins regularizer to align cross-modal representations without negative samples. Experiments on three public datasets show that CIG-MAE consistently outperforms SOTA SSL methods and even surpasses a fully supervised baseline, demonstrating superior data efficiency, robustness, and representation generalization.
Read the full abstract

Computer Vision · Cross-listed from cs.CC

Hardware-Algorithm Co-Optimization of Early-Exit Neural Networks for Multi-Core Edge Accelerators

The deployment of Early-Exiting Neural Networks (EENNs) on edge accelerators requires optimizing not only the network architecture but also its hardware deployment. Exit configuration, quantization, and hardware workload mapping interact in non-trivial ways, influencing memory traffic, accelerator utilization, and ultimately the energy-latency trade-off. This work presents a hardware-aware co-design framework for EENNs that jointly optimizes exit configuration, quantization-aware training, and multi-core hardware mapping within a unified NAS process. Leveraging analytical design space exploration, the framework identifies efficient workload mappings for each candidate architecture while providing accurate latency and energy estimates during the search. We further formulate EENN deployment as a constrained multi-objective optimization problem balancing predictive accuracy, energy-latency product, exit overhead, and dynamic inference efficiency. Experimental results on CIFAR-10 demonstrate that the proposed framework achieves over a 50% reduction in energy-latency product compared with static baselines under 8-bit quantization. These results demonstrate that jointly optimizing architecture and deployment is essential for realizing the full efficiency potential of dynamic inference on heterogeneous edge accelerators.
Read the full abstract

All authors Alaa Zniber, Arne Symons, Ouassim Karrakchou, Marian Verhelst, Mounir Ghogho

Computer Vision

OmniScaleSR: Unleashing Scale-Controlled Diffusion Prior for Faithful and Realistic Arbitrary-Scale Image Super-Resolution

Arbitrary-scale super-resolution (ASSR) overcomes the limitation of traditional super-resolution (SR) methods that operate only at fixed scales (e.g., 4x), enabling a single model to handle arbitrary magnification. Most existing ASSR approaches rely on implicit neural representation (INR), but its regression-driven feature extraction and aggregation intrinsically limit the ability to synthesize fine details, leading to low realism. Recent diffusion-based realistic image super-resolution (Real-ISR) models leverage powerful pre-trained diffusion priors and show impressive results at the 4x setting. We observe that they can also achieve ASSR because the diffusion prior implicitly adapts to scale by encouraging high-realism generation. However, without explicit scale control, the diffusion process cannot be properly adjusted for different magnification levels, resulting in excessive hallucination or blurry outputs, especially under ultra-high scales. To address these issues, we propose OmniScaleSR, a diffusion-based realistic arbitrary-scale SR framework designed to achieve both high fidelity and high realism. We introduce explicit, diffusion-native scale control mechanisms that work synergistically with implicit scale adaptation, enabling scale-aware and content-aware modulation of the diffusion process. In addition, we incorporate multi-domain fidelity enhancement designs to further improve reconstruction accuracy. Extensive experiments on bicubic degradation benchmarks and real-world datasets show that OmniScaleSR surpasses state-of-the-art methods in both fidelity and perceptual realism, with particularly strong performance at large magnification factors. Code will be released at https://github.com/chaixinning/OmniScaleSR.
Read the full abstract

All authors Xinning Chai, Zhengxue Cheng, Yuhong Zhang, Hengsheng Zhang, Yingsheng Qin, Yucai Yang, Rong Xie, Li Song

Computer Vision

Towards Cross-View Point Correspondence in Vision-Language Models

Cross-view correspondence is a fundamental capability for spatial understanding and embodied AI. However, it is still far from being realized in Vision-Language Models (VLMs), especially in achieving precise point-level correspondence, which is crucial for precise affordance interaction. So we propose the Cross-View Point Correspondence (CVPC) task and CrossPoint-Bench, a comprehensive benchmark with hierarchical design, inspired by the human cognitive process of "perceive", "reason", and "correspond". Our evaluation shows the state-of-the-art models (e.g., Gemini-2.5-Pro) still fall far behind humans, with a gap of over 54.65% in overall accuracy, exposing a challenge in transitioning from coarse-grained judgement to fine-grained coordinate prediction. To address this problem, we construct CrossPoint-378K, a dataset with 378K question-answering pairs across 900 scenes, focused on actionable affordance regions that better reflect real-world manipulation and interaction scenarios. Furthermore, we propose CroPond that trained on the CrossPoint-378K dataset. Our CroPond achieves state-of-the-art performance on CrossPoint-Bench, surpassing Gemini-2.5-Pro by 39.7% accuracy, which offers a foundation for advancing future work on cross-view correspondence. The benchmark, dataset, and model are publicly available at https://github.com/WangYipu2002/CrossPoint.
Read the full abstract

All authors Yipu Wang, Yuheng Ji, Yuyang Liu, Enshen Zhou, Ziqiang Yang, Yuxuan Tian, Ziheng Qin, Yue Liu, Huajie Tan, Cheng Chi, Zhiyuan Ma, Daniel Dajun Zeng, Xiaolong Zheng

Computer Vision

Reward Forcing: Efficient Streaming Video Generation with Rewarded Distribution Matching Distillation

Efficient streaming video generation is critical for simulating interactive and dynamic worlds. Existing methods distill few-step video diffusion models with sliding window attention, using initial frames as sink tokens to maintain attention performance and reduce error accumulation. However, video frames become overly dependent on these static tokens, resulting in copied initial frames and diminished motion dynamics. To address this, we introduce Reward Forcing, a novel framework with two key designs. First, we propose EMA-Sink, which maintains fixed-size tokens initialized from initial frames and continuously updated by fusing evicted tokens via exponential moving average as they exit the sliding window. Without additional computation cost, EMA-Sink tokens capture both long-term context and recent dynamics, preventing initial frame copying while maintaining long-horizon consistency. Second, to better distill motion dynamics from teacher models, we propose a novel Rewarded Distribution Matching Distillation (Re-DMD). Vanilla distribution matching treats every training sample equally, limiting the model's ability to prioritize dynamic content. Instead, Re-DMD biases the model's output distribution toward high-reward regions by prioritizing samples with greater dynamics rated by a vision-language model. Re-DMD significantly enhances motion quality while preserving data fidelity. We include both quantitative and qualitative experiments to show that Reward Forcing achieves state-of-the-art performance on standard benchmarks while enabling high-quality streaming video generation at 23.1 FPS on a single H100 GPU.
Read the full abstract

All authors Yunhong Lu, Yanhong Zeng, Haobo Li, Hao Ouyang, Qiuyu Wang, Ka Leong Cheng, Jiapeng Zhu, Hengyuan Cao, Zhipeng Zhang, Xing Zhu, Yujun Shen, Min Zhang

Computer Vision

Live Avatar: Streaming Real-time Audio-Driven Avatar Generation with Infinite Length

Audio-driven avatar interaction demands real-time, streaming, and infinite-length generation – capabilities fundamentally at odds with the sequential denoising and long-horizon drift of current diffusion models. We present Live Avatar, an algorithm-system co-designed framework that addresses both challenges for a 14-billion-parameter diffusion model. On the algorithm side, a two-stage pipeline distills a pretrained bidirectional model into a causal, few-step streaming one, while a set of complementary long-horizon strategies eliminate identity drift and visual artifacts, enabling stable autoregressive generation exceeding 10000 seconds. On the system side, Timestep-forcing Pipeline Parallelism (TPP) assigns each GPU a fixed denoising timestep, converting the sequential diffusion chain into an asynchronous spatial pipeline that simultaneously boosts throughput and improves temporal consistency. Live Avatar achieves 45 FPS with a TTFF of 1.21\,s on 5 H800 GPUs, and to our knowledge is the first to enable practical real-time streaming of a 14B diffusion model for infinite-length avatar generation. We further introduce GenBench, a standardized long-form benchmark, to facilitate reproducible evaluation. Our project page is at https://liveavatar.github.io/.
Read the full abstract

All authors Yubo Huang, Hailong Guo, Fangtai Wu, Weiqiang Wang, Shifeng Zhang, Shijie Huang, Qijun Gan, Lin Liu, Sirui Zhao, Enhong Chen, Jiaming Liu, Steven Hoi

Computer Vision

I2I-Bench: A Comprehensive Benchmark Suite for Image-to-Image Editing Models

Image editing models are advancing rapidly, yet comprehensive evaluation remains a significant challenge. Existing image editing benchmarks generally suffer from limited task scopes, insufficient evaluation dimensions, and heavy reliance on manual annotations, which significantly constrain their scalability and practical applicability. To address this, we propose I2I-Bench, a comprehensive benchmark for image-to-image editing models, which features (i) diverse tasks, encompassing 10 task categories across both single-image and multi-image editing tasks, (ii) comprehensive evaluation dimensions, including 30 decoupled and fine-grained evaluation dimensions with automated hybrid evaluation methods that integrate specialized tools and large multimodal models (LMMs), and (iii) rigorous alignment validation, justifying the consistency between our benchmark evaluations and human preferences. Using I2I-Bench, we benchmark numerous mainstream image editing models, investigating the gaps and trade-offs between editing models across various dimensions. We will open-source all components of I2I-Bench to facilitate future research.
Read the full abstract

All authors Juntong Wang, Jiarui Wang, Huiyu Duan, Jiaxiang Kang, Guangtao Zhai, Xiongkuo Min

Computer Vision

SEASON: Mitigating Temporal Hallucination in Video Large Language Models via Self-Diagnostic Contrastive Decoding

Video Large Language Models (VideoLLMs) have shown remarkable progress in video understanding. However, these models still struggle to effectively perceive and exploit rich temporal information in videos when responding to user queries. Therefore, they often generate descriptions of events that are temporal inconsistent or causally implausible, causing severe hallucination issues. While most prior studies have focused on spatial hallucinations (e.g. object mismatches), temporal reasoning in video understanding remains relatively underexplored. To address this issue, we propose Self-Diagnostic Contrastive Decoding (SEASON), a training-free method that adaptively enhances temporal and spatial faithfulness for each output token. It achieves this by dynamically diagnosing each token's hallucination tendency and applying adaptive contrastive decoding against its corresponding temporal and spatial negatives. Extensive experiments demonstrate that SEASON outperforms all existing training-free hallucination mitigation approaches on three hallucination examination benchmarks, while further improves VideoLLMs across four general video understanding benchmarks. The code will be released upon acceptance.
Read the full abstract

All authors Chang-Hsun Wu, Kai-Po Chang, Yu-Yang Sheng, Hung-Kai Chung, Kuei-Chun Wang, Yu-Chiang Frank Wang

Computer Vision · Cross-listed from cs.LG

Rethinking Decoupled Knowledge Distillation: A Predictive Distribution Perspective

In the history of knowledge distillation, the focus has once shifted over time from logit-based to feature-based approaches. However, this transition has been revisited with the advent of Decoupled Knowledge Distillation (DKD), which re-emphasizes the importance of logit knowledge through advanced decoupling and weighting strategies. While DKD marks a significant advancement, its underlying mechanisms merit deeper exploration. As a response, we rethink DKD from a predictive distribution perspective. First, we introduce an enhanced version, the Generalized Decoupled Knowledge Distillation (GDKD) loss, which offers a more versatile method for decoupling logits. Then we pay particular attention to the teacher model's predictive distribution and its impact on the gradients of GDKD loss, uncovering two critical insights often overlooked: (1) the partitioning by the top logit considerably improves the interrelationship of non-top logits, and (2) amplifying the focus on the distillation loss of non-top logits enhances the knowledge extraction among them. Utilizing these insights, we further propose a streamlined GDKD algorithm with an efficient partition strategy to handle the multimodality of teacher models' predictive distribution. Our comprehensive experiments conducted on a variety of benchmarks, including CIFAR-100, ImageNet, Tiny-ImageNet, CUB-200-2011, and Cityscapes, demonstrate GDKD's superior performance over both the original DKD and other leading knowledge distillation methods. The code is available at https://github.com/ZaberKo/GDKD.
Read the full abstract

Computer Vision

Denoise to Track: Harnessing Video Diffusion Priors for Robust Correspondence

In this work, we introduce HeFT (Head-Frequency Tracker), a zero-shot point tracking framework that leverages the visual priors of pretrained video diffusion models. To better understand how they encode spatiotemporal information, we analyze the internal representations of Video Diffusion Transformer (VDiT). Our analysis reveals that attention heads act as minimal functional units with distinct specializations for matching, semantic understanding, and positional encoding. Additionally, we find that the low-frequency components in VDiT features are crucial for establishing correspondences, whereas the high-frequency components tend to introduce noise. Building on these insights, we propose a head- and frequency-aware feature selection strategy that jointly selects the most informative attention head and low-frequency components to enhance tracking performance. Specifically, our method extracts discriminative features through single-step denoising, applies feature selection, and employs soft-argmax localization with forward-backward consistency checks for correspondence estimation. Extensive experiments on TAP-Vid benchmarks demonstrate that HeFT achieves state-of-the-art zero-shot tracking performance, approaching the accuracy of supervised methods while eliminating the need for annotated training data. Our work further underscores the promise of video diffusion models as powerful foundation models for a wide range of downstream tasks, paving the way toward unified visual foundation models.
Read the full abstract

All authors Tianyu Yuan, Yuanbo Yang, Lin-Zhuo Chen, Yao Yao, Zhuzhong Qian

Computer Vision

Malicious Image Analysis via Vision-Language Segmentation Fusion: Detection, Element, and Location in One-shot

Detecting illicit visual content demands more than image-level NSFW flags; moderators must also know what objects make an image illegal and where those objects occur. We introduce a zero-shot pipeline that simultaneously (i) detects if an image contains harmful content, (ii) identifies each critical element involved, and (iii) localizes those elements with pixel-accurate masks - all in one pass. The system first applies foundation segmentation model (SAM) to generate candidate object masks and refines them into larger independent regions. Each region is scored for malicious relevance by a vision-language model using open-vocabulary prompts; these scores weight a fusion step that produces a consolidated malicious object map. An ensemble across multiple segmenters hardens the pipeline against adaptive attacks that target any single segmentation method. Evaluated on a newly-annotated 790-image dataset spanning drug, sexual, violent and extremist content, our method attains 85.8% element-level recall, 78.1% precision and a 92.1% segment-success rate - exceeding direct zero-shot VLM localization by 27.4% recall at comparable precision. Against PGD adversarial perturbations crafted to break SAM and VLM, our method's precision and recall decreased by no more than 10%, demonstrating high robustness against attacks. The full pipeline processes an image in seconds, plugs seamlessly into existing VLM workflows, and constitutes the first practical tool for fine-grained, explainable malicious-image moderation.
Read the full abstract

All authors Sheng Hang, Chaoxiang He, Hongsheng Hu, Hanqing Hu, Bin Benjamin Zhu, Shi-Feng Sun, Dawu Gu, Shuo Wang

Computer Vision

When Robots Should Say "I Don't Know": Benchmarking Abstention in Embodied Question Answering

Embodied Question Answering (EQA) requires an agent to interpret language, perceive its environment, and navigate within 3D scenes to produce responses. Existing EQA benchmarks assume that every question must be answered, but embodied agents should know when they do not have sufficient information to answer. In this work, we focus on a minimal requirement for EQA agents, abstention: knowing when to withhold an answer. From an initial study of 500 human queries, we find that 32.4% contain missing or underspecified context. Drawing on this initial study and cognitive theories of human communication errors, we derive five representative categories requiring abstention: actionability limitation, referential underspecification, preference dependence, information unavailability, and false presupposition. We augment OpenEQA by having annotators transform well-posed questions into ambiguous variants outlined by these categories. The resulting dataset, AbstainEQA, comprises 1,636 annotated abstention cases paired with 1,636 original OpenEQA instances for balanced evaluation. Evaluating on AbstainEQA, we find that even the best frontier model only attains 42.79% abstention recall, while humans achieve 91.17%. We also find that scaling, prompting, and reasoning only yield marginal gains, and that fine-tuned models overfit to textual cues. Together, these results position abstention as a fundamental prerequisite for reliable interaction in embodied settings and as a necessary basis for effective clarification.
Read the full abstract

All authors Tao Wu, Chuhao Zhou, Guangyu Zhao, Haozhi Cao, Yewen Pu, Jianfei Yang

Computer Vision

SAM3-I: Segment Anything with Instructions

Segment Anything Model 3 (SAM3) advances open-vocabulary segmentation through promptable concept segmentation, enabling users to segment all instances associated with a given concept using short noun-phrase (NP) prompts. While effective for concept-level grounding, real-world interactions often involve far richer natural-language instructions that combine attributes, relations, actions, states, or implicit reasoning. Currently, SAM3 relies on external multi-modal agents to convert complex instructions into NPs and conducts iterative mask filtering, leading to coarse representations and limited instance specificity. In this work, we present SAM3-I, an instruction-following extension of the SAM family that unifies concept-level grounding and instruction-level reasoning within a single segmentation framework. Built upon SAM3, SAM3-I introduces an instruction-aware cascaded adaptation mechanism with dedicated alignment losses that progressively aligns expressive instruction semantics with SAM3's vision-language representations, enabling direct interpretation of natural-language instructions while preserving its strong concept recall ability. To enable instruction-following learning, we introduce HMPL-Instruct, a large-scale instruction-centric dataset that systematically covers hierarchical instruction semantics and diverse target granularities. Experiments demonstrate that SAM3-I achieves appealing performance across referring and reasoning-based segmentation, showing that SAM3 can be effectively extended to follow complex natural-language instructions without sacrificing its original concept-driven strengths. Code and dataset are available at https://github.com/debby-0527/SAM3-I.
Read the full abstract

All authors Jingjing Li, Yue Feng, Yuchen Guo, Jincai Huang, Wei Ji, Qi Bi, Yongri Piao, Miao Zhang, Xiaoqi Zhao, Qiang Chen, Shihao Zou, Huchuan Lu, Li Cheng

Computer Vision

Infrared UAV Target Tracking with Dynamic Feature Refinement and Global Contextual Attention Knowledge Distillation

Unmanned aerial vehicle (UAV) target tracking based on thermal infrared imaging has been one of the most important sensing technologies in anti-UAV applications. However, the infrared UAV targets often exhibit weak features and complex backgrounds, posing significant challenges to accurate tracking. To address these problems, we introduce SiamDFF, a novel dynamic feature fusion Siamese network that integrates feature enhancement and global contextual attention knowledge distillation for infrared UAV target (IRUT) tracking. The SiamDFF incorporates a selective target enhancement network (STEN), a dynamic spatial feature aggregation module (DSFAM), and a dynamic channel feature aggregation module (DCFAM). The STEN employs intensity-aware multi-head cross-attention to adaptively enhance important regions for both template and search branches. The DSFAM enhances multi-scale UAV target features by integrating local details with global features, utilizing spatial attention guidance within the search frame. The DCFAM effectively integrates the mixed template generated from STEN in the template branch and original template, avoiding excessive background interference with the template and thereby enhancing the emphasis on UAV target region features within the search frame. Furthermore, to enhance the feature extraction capabilities of the network for IRUT without adding extra computational burden, we propose a novel tracking-specific target-aware contextual attention knowledge distiller. It transfers the target prior from the teacher network to the student model, significantly improving the student network's focus on informative regions at each hierarchical level of the backbone network. Extensive experiments on real infrared UAV datasets demonstrate that the proposed approach outperforms state-of-the-art target trackers under complex backgrounds while achieving a real-time tracking speed.
Read the full abstract

All authors Houzhang Fang, Chenxing Wu, Kun Bai, Tianqi Chen, Xiaolin Wang, Xiyang Liu, Yi Chang, Luxin Yan